Ellen Meiksins Wood
Rational Choice Marxism: Is the Game Worth the Candle?
Some time ago, in the pages of the New Left Review, a claim was made on behalf of ‘rational-choice Marxism’ as ‘a fully fledged paradigm, which deserves to take its place beside the two other constellations of theory currently discernible within the broad spectrum ofprogressive social thought—namely, post-structuralism and critical theory’. [*] I am very grateful to Perry Anderson, Robert Brenner, Diane Elson, Norman Geras and Neal Wood for their comments and suggestions.  Alan Carling, ‘Rational Choice Marxism’, nlr 160, November/December 1986, p. 55. (Hereafter, this article will be cited as Carling I.) More than that: ‘it is now only within the rational-choice context that some of the leading items on the classical agenda of Marxist theory—historical explanation and the delineation of social form, the collective dynamics of class struggle, the evolution and evaluation of capitalism—can be fruitfully discussed.’ These are very large claims, and if this new ‘paradigm’ can even partially live up to them, it deserves the vogue it is now enjoying in the Anglo-American academy. A theoretical advance in any one of the ‘leading items on the classical agenda of Marxist theory’ would be a worthy accomplishment; but it would indeed be a remarkable achievement if, even without driving any likely competitors from the field, this body of thought could be shown to merit the status of a ‘fully fledged paradigm’, a comprehensive theoretical ‘constellation’ with the explanatory range of classical Marxism.
1. What is Rational Choice Marxism?
There is a difficulty at the outset in evaluating the claims of rcm. What are to be our criteria of inclusion? Alan Carling tells us that ‘if there is one distinctive presupposition of the intended body of work, it must be the view that societies are composed of human individuals who, being endowed with resources of various kinds, attempt to choose rationally between various courses of action.’  Carling I, pp. 26–7. On the face of it, this ‘distinction’ places rcm in the company of a dauntingly large and heterogeneous collection of writers—only part of which is exhausted by Carling’s own admission that this same ‘distinctive presupposition’ is ‘a commonplace of that broad sweep of conventional economics and philosophy conducted in the liberal tradition.’ The further specification that the ‘dramatic’ difference lies in rcm’s ‘joining of this presupposition with the classical agenda of Marxist theory’ only complicates matters. To guarantee the comprehensiveness of this paradigm, its full coverage of the classical Marxist agenda, Carling is obliged to collect under its rubric a wide and disparate range of writers, at least some of whom would contest their own inclusion (he specifically acknowledges Norman Geras’s disavowal of the rcm label, though without apparently accepting this self-exclusion). It is not enough to include figures like John Roemer or Jon Elster, those who would most readily accept a ‘rational-choice’ designation which carries the associations of game theory and methodological individualism. To make good his claims for rcm in the areas of historical explanation in general and the evolution of capitalism in particular—themes which on any conventional reckoning are central to the Marxist agenda—Carling must recruit into the rcm school writers who evince a very different attitude to methodological individualism—notably Robert Brenner and G.A. Cohen.
In other words, the ‘distinctive presupposition’ must remain very broadly defined indeed (so broadly, in fact, that it is difficult to see why that most ‘classical’ of Marxists, Friedrich Engels—who, after all, had something to say about the formation of social patterns from the unintended consequences of individual human actions—could not be taken on board). To say that this paradigm is also distinctively characterized by an ‘analytic’ mode of presentation (the term ‘analytic Marxism’ is sometimes taken to be synonymous with rational-choice Marxism) hardly advances matters, since the ‘analytic’ style of argumentation is in principle compatible with any set of substantive propositions. Alternatively, if the ‘distinctive presupposition’ is further specified, if rcm is distinguished by its game-theoretic methodological individualism, then the most distinctive characteristic of this paradigm—the formal abstraction and the static, ahistorical individualism of the rational choice model—is the one which is least congenial to those central items of the Marxist agenda: social change and historical process, and in particular, the ‘laws of motion’ specific to various modes of production, their characteristic crises, the specific principles of motion in the transition from one mode to another.
Carling himself seems aware of the difficulty when he acknowledges that Roemer’s theory of exploitation provides no substitute of its own for the ‘classic’ Marxist theory of history but remains ‘parasitic’ on that general theory (as interpreted by G.A. Cohen).  Carling I, p. 52. Indeed, Roemer makes no pretence that his distinctive rational-choice paradigm, which he offers as a substitute for the classical theory of exploitation based on the labour theory of value, can generate its own theory of historical process. What he offers us instead is a hybrid. As Carling puts it in his recent review of Roemer’s latest book Free to Lose (‘the textbook of contemporary Marxist theory’  Carling, ‘Liberty, Equality, Community’, nlr 171, September/October 1988, p. 95, (Hereafter, Carling II.)), Roemer’s ‘standard contemporary Marxism’ is ‘Cohen on history plus Roemer on class and exploitation’.
A sensible way to proceed, then, might be to begin with a specific definition of rcm, which recognizes the distinctiveness of its gametheoretic models and their methodological individualism, but without prejudging the capacity of this paradigm to enter into a fruitful alliance with a theory of social change and historical process from outside its distinctive methodological boundaries. This leaves us, in the first instance, with three obvious major candidates for inclusion: John Roemer, Jon Elster, and (lately) Adam Przeworski. The latter is included with some hesitation, because, although he sometimes carries the rational-choice model to extremes not envisaged even by Roemer, he is the least consistent of the three in his adherence to the paradigm, having done important political analysis which does not rely on the game-theoretic paradigm and which is, if anything, undermined by the model’s theoretical demands. In what follows, the focus will be mainly on Roemer, who has provided the clearest and most comprehensive account of the rcm paradigm, with excursions into Elster and Przeworski when they help to clarify some important point or explicitly depart from Roemer in some significant way. Erik Olin Wright will be excluded from consideration simply on the grounds that his theoretical formation appears to be in a critical stage of transition: his recent book Classes, which constructed a theory of class on the foundations of Roemer’s theory of exploitation, was almost immediately followed by an article co-authored with Andrew Levine and Elliott Sober which deflates the explanatory pretensions of methodological individualism practically to the vanishing point.  Andrew Levine, Elliott Sober, and Erik Olin Wright, ‘Marxism and Methodological Individualism’, nlr 162, March/April 1987, pp. 67–84. For an even more recent version of Wright’s theory of class, see Wright et al., The Debate on Classes, Verso, orthcoming. The connection of Brenner and Cohen to this core group remains to be established and will be explored when we come to the question of rcm’s treatment of history.
Roemer has written that ‘methodological individualism’, specifically in its ‘game-theoretic’ mode, is essential to the explication of historical materialism and its ‘key propositions’: ‘the key questions of historical materialism,’ he maintains, ‘require reference to the specific forms of class struggle, and . . . an understanding of such struggles is elucidated by game theory; . . . class analysis requires microfoundations at the level of the individual to explain why and when classes are the relevant unit of analysis.’  John Roemer, ‘Methodological Individualism and Deductive Marxism’, Theory and Society 11 (1982), p. 513. Let us say, then, that the differentia specifica of rcm is a theory of exploitation and class in which various modes of exploitation and classes are logically generated according to the principles of game-theoretic analysis. This by no means reduces rcm to a purely methodological strategy. On the contrary, as we shall see, the form of the rcm theory is to a large extent its substance, and in its gametheoretic assumptions are secreted vital substantive theses about the social world.
The Roots of Rational Choice
The elusiveness of rcm’s distinctive identity may have something to do with a certain lack of critical self-awareness among its practitioners. They are remarkably insensitive to the history and context of ideas in general, and their own ideas in particular; and they are generally inclined to remain within a very narrow universe of debate. rcmists, to judge by the extent of their mutually self-confirming footnoting and the restricted range of controversy encompassed by their writings, seem to talk largely to each other.
Despite its characteristic ahistoricism, the provenance of the paradigm is clear enough. The roots of this game-theoretic rational-choice approach to social theory are to be found in conventional neo-classical economics and its extension to the other ‘social sciences’ in the work of such writers as James Buchanan, Anthony Downs, Mancur Olson, and Gary Becker. In other words, the ‘rational choice’ paradigm has its origins in the rebirth of right-wing thought. This is not to say that the theoretical associations of the paradigm must inevitably propel its adherents to the political right, but since this filiation is never consciously and critically confronted, the resistance of rcm’s political impulses to attractions from the right is seriously weakened.
There are also, however, other intellectual contexts, less alien to Marxism, which might help to explain the emergence of rcm. The dominant school of Marxism during the period of rcm gestation was undoubtedly Althusserian structuralism. It is perhaps against the background of this Althusserian hegemony, and the excesses of its attacks on conceptions of human agency in favour of structural explanations from which the human subject was ‘rigorously’ expelled, that the attractions of the methodological individualism proposed by Roemer and Elster can be most sympathetically understood—though, as we shall see, there have been some unexpected points of convergence between these two apparently antithetical paradigms.
One more intellectual co-ordinate is worth mentioning: the analytic, formalistic mode of academic political philosophy, both liberal and conservative, which has evolved especially in the United States, as exemplified in particular by John Rawls and Robert Nozick. It is possible that rcm can most positively be conceived as an attempt to construct a normative socialist theory to counter the conservative philosophy of writers like Nozick. To say this, however, may be to circumscribe the claims which rcm makes for itself, to limit its application far more strictly than any of its exponents have yet been willing to do. Although Roemer in particular maintains that a principal benefit of the rational choice paradigm is that it reveals, as conventional Marxist theory cannot do, the ethical basis of Marxism, the claims for rcm as a superior explanatory model have been equally vigorous.
2. The rcm Theory of Exploitation and Class
The locus classicus of the rcm theory of class is John Roemer’s ‘general theory’ of exploitation. Jon Elster has described Roemer’s methodological individualism as ‘generating class relations and the capital relationship from exchanges between differently endowed individuals in a competitive setting’.  Jon Elster, Making Sense of Marx, Cambridge 1985, p. 7. More generally, Roemer’s approach is an attempt to identify the ‘key moment’ in class exploitation by taking as his point of departure nothing more than individuals possessed of different endowments and showing how unequal distribution of the relevant assets necessarily produces unfair results, especially in the distribution of income, even without a direct transfer of labour from exploiter to exploited.
Beginning with a ‘game-theoretic’ model which classifies exploitative systems according to the various resources or assets that are respectively salient in each one, he concludes that people are exploited when they would be better off if the relevant asset were (according to his ‘withdrawal’ criterion  According to this criterion, a ‘coalition of agents’ is exploited if it would be better off withdrawing from, rather than remaining within, the ‘game’ with its per capita share of whatever assets are relevant to the mode of exploitation in question.) differently allocated, and that they are exploiters when a reallocation of the relevant asset would leave them worse off. This definition, in accordance with the principles of game theory, requires that each mode of exploitation be defined in terms of a ‘hypothetically feasible’ alternative, establishing the criterion against which the status of exploiter and exploited can be assessed. So, for example, feudal exploitation entails ‘differential access to freedom from bondage’; that is, ‘feudal exploitation is that inequality which arises as a consequence of ties of bondage which prevent producers from freely engaging in trade, with their own assets.’  Roemer, A General Theory of Exploitation and Class, Cambridge, Mass. 1982, pp. 20–21. Here, the ‘hypothetical alternative’ is capitalism; and the test of feudal exploitation ‘is to calculate whether a person would be better off under the distribution where no feudal property exists but capitalist property still exists.’  Roemer, Free to Lose, London 1988, p. 136.
The essential feature of this theory is its focus on what Roemer calls ‘property relations’. What he means by ‘property relations’ is the distribution of assets or endowments, not the social relations of production and appropriation as Marxism commonly understands them. Indeed, it is the object of this focus on ‘property relations’ to shift the criterion of exploitation away from the direct relations between producers and appropriators and locate it in distributional factors. Thus the traditional Marxist concept of surplus extraction is rejected, while the relations and processes it entails are displaced by what might be called the static and indirect ‘relations’ of relative advantage.
The purpose of this reconceptualization is, in particular, to explain capitalist exploitation without resorting to the labour theory of value and the concept of surplus value, which Roemer judges to be inadequate as a measure of exploitation. But he does not stop here, with the ‘special’ theory of capitalism. In order to provide a general theory under which the special theory of capitalism can be subsumed and which can also encompass socialism, he jettisons not only the concept of surplus value specific to capitalism but the whole notion of exploitation as the extraction of surplus labour. Unequal distribution of assets, rather than the relations between direct producers and the appropriators of their surplus labour, becomes the central focus.
This distributional theory of exploitation marks a critical departure from ‘classic’ Marxism. For conventional Marxism, inequality has no theoretical purchase except insofar as it entails a system of social relations between appropriators and producers. In those relations, and not in inequality as such, lies the dynamic principle, the contradictions and conflicts, which account for social and historical processes. The project of rcm, in direct opposition to traditional Marxism, is to construct a conception of exploitation which does not require reference to such relations. Where there is a direct link, a class relation, between exploiter and exploited—and not just a ‘relation’ of relative advantage—it is established by ‘rational choice’.
Roemer’s principal claim for his theory and its advantages over classical Marxism is that it can ‘finesse all the criticisms of Marxian exploitation which can be made because of its reliance on the labour theory of value.’  Roemer, General Theory, p. 20. In particular, he maintains that his ‘game-theoretic formulation is superior in that it makes explicit what ethical presumption lies behind the Marxian theory of exploitation’, by taking as its point of departure the alternative distribution of relevant assets against which relative advantages and disadvantages can be measured. It is, in fact, largely for the purpose of moral argumentation that he adopts his characteristic paradigm, and it is perhaps here, in its contribution to ethical arguments about exploitation, that rcm claims have their greatest validity. But since Roemer has sought to reconstruct the whole of Marxist thought on the basis of his game-theoretic theory of exploitation, we need to consider first the explanatory implications of the initial decision to displace the relations of surplus extraction from their central position in the theory of exploitation.
The Explanatory Consequences
Let us assume, for the sake of argument, that Roemer has succeeded on his own terms in constructing a foolproof demonstration that capitalism is exploitative and unjust, without resorting to the labour theory of value.  This is certainly not to deny that Roemer’s theory of exploitation is open to criticism even on its own economic terms. See, for example, Ronald A. Kieve, ‘From Necessary Illusion to Rational Choice? A Critique of Neo-Marxist Rational-Choice Theory’, Theory and Society 15 (1986), esp. pp. 558–565. My own argument has more to do with evaluating the utility of this theory even if it succeeds on its own terms. How does this procedure affect the larger project of reconstructing Marxist theory? Here it is important to keep in mind that the idea of surplus labour is not principally (or probably at all) intended to provide the kind of mathematical measure which Roemer demands for his moral indictment of exploitation in general and capitalism in particular. In any case, whatever the weaknesses of the concept as a mathematical device or even as a moral standard, they are not relevant to the one fundamental insight encased in this controversial idea: that at the foundation of social life and historical processes lie the conditions in which people have access to the means of subsistence and reproduction, and that a decisive historical break occurs when the prevailing conditions systematically compel some people to transfer part of their labour or its product to someone else. The critical datum in explaining social processes is not the quantitative measure of the ‘surplus’ but rather the very fact of the relationship between producers and appropriators, and the conditions in which it occurs; and this critical fact can be established without mathematical proof. Indeed, Roemer himself must begin with a tacit recognition of this fact, in order to have something to measure.
Capitalism undoubtedly represents a special case, because capitalist appropriation is not a distinctly visible act—like, say, the serf’s payment of dues to the lord—which constitutes a separate act of appropriation, after the fact of the serf’s labour and in the context of a visible relation between appropriator and producer. In contrast, there is no immediately obvious way of separating the act of capitalist appropriation from the process of production or from the process of commodity exchange through which capital realizes its gains. The concept of surplus value is meant to convey this complex relation between production, realization in commodity exchange, and capitalist appropriation.
There has been no shortage of critics, including Marxist economists, who are keen to point out the difficulties of expressing these relations in quantitative terms—that is, of measuring ‘value’ and ‘surplus’, or of relating ‘value’ to ‘price’. But Roemer has done more than simply jettison an imperfect standard of measurement. He has elaborated a theory of capitalist exploitation which displaces the relevant social relations themselves from the centre of Marxist analysis. This may or may not be necessary to a persuasive moral argument against capitalism, but in either case, it remains to be seen whether the explanatory price is worth paying.
What, then, are the consequences of conceiving class exploitation as a ‘relation’ of relative advantage, or at best a ‘rational’ exchange between individuals, a conception in which the analytic starting point is inequality or the ‘unequal distribution of assets’, instead of a (historically constituted) social relation between appropriators and producers?
First, the individual and his/her ‘endowments’: the purpose of the rcm exercise is to break down ‘macroprocesses’ into their ‘microfoundations’, the actions of individuals. But it turns out that rcm can achieve its objective of ‘beginning’ with the individual only by a sleight of hand; while the only direct relation overtly admitted by the theory is the ‘rationally chosen’ exchange between individuals, whole sets of relations or structures must be covertly secreted in the ‘differential endowments’ which constitute the individuals in question.
For example, in Roemer’s definition of feudal exploitation, ‘bondage’—with all the relations of power, domination and juridical dependence between lords and peasants which it entails—is treated as an attribute of individuals, what has been described as a ‘relational property’. How, after all, can ‘bondage’ or ‘freedom’ be conceived as an individual attribute, in abstraction from relations with others? Hidden within the individual peasant’s ‘assets’ are not only his possessions, but all the structural relations which render those possessions relevant, which, in other words, constitute them—or some of them—as ‘assets’ at all; the communal organization that maintains peasant possession in their means of subsistence; and so on. Similarly, the lord’s ‘assets’ encompass, among other things, his location in a structure of relations which make him capable of exerting extra-economic coercion; a community of lords, a state-formation, etc. In fact, it would be impossible to characterize the relevant ‘assets’ without first specifying the whole system of social relations that constitute them as assets in the first place for any given individual.
Analogous problems arise in dealing with capitalism. Methodological individualism requires that the behaviour of capitalists be reducible to their individual assets and motivations, on the basis of information visible to them. (Jon Elster, the most stringent critic of Marx for his so-called ‘methodological collectivism’, attacks Marx’s value theory, for example, on the grounds that ‘individual behaviour can never be explained by reference to values, which, being invisible, have no place in the purposive explanation of action.’  Elster, p. 515.) Yet the compulsions of capital accumulation cannot be derived simply from the ‘optimizing strategies’ of a rational individual with capital ‘assets’. Those compulsions cannot be explained without reference to the competitive pressures of the capitalist market, indeed the whole historically constituted social structure which has made individuals in capitalist society uniquely dependent on the market for the conditions of their selfreproduction, and hence subject to the imperatives of competition and accumulation.
Having repudiated the conceptual apparatus by means of which ‘classic’ Marxism seeks to explain these systemic imperatives, rcm is obliged to take as given the compulsions of capitalism, and to impute them to the preferences and motivations of individual capitalists. It is no longer simply a matter of beginning with certain general assumptions about human nature; we must now take as our starting point the specific attributes of the capitalist nature. Indeed, it is difficult to see how the rcm model can escape a complete circularity, according to which individuals accumulate capital because they are capital accumulators. The impulse to accumulate capital itself cannot be further reduced to individual properties independent of social structures. Instead, the properties of capitalism as a social system, its systemic compulsion to accumulate, and arguably even the theory of value itself—everything that Marx sought to explain rather than assume—must be simply incorporated, taken for granted and not explained, into the individual ‘properties’ of the capitalist. And even then, since the individual capitalist cannot accommodate among his/her personal attributes all the complex transactions of capitalism which make accumulation possible as well as necessary, nor all the conditions which consistently upset the assumptions on which accumulation is based, the explanatory power of the rcm model is rather limited—without the constant intrusion of unexplained assumptions, or constant assistance from conventional Marxism. Whatever capacity the model has to clarify the issues of capital accumulation—by explicating the motivations of capitalists only on the basis of the information which is visible to them—depends on assuming first what most needs to be explained: the logic of the capitalist system which imposes the compulsion of accumulation.
At best, this means that all the important work must be done in advance of applying the rcm model, and entirely without its help. All the historical analysis which yields the structures that are to be smuggled into the attributes of the relevant individuals must be done first, very likely with the tools of conventional historical materialism; or, more precisely, the rcm model must take as given precisely what needs to be explained. Indeed, the individual ‘properties’ which motivate the ‘rational choice’ must in effect be deduced from the ‘macroprocesses’ which are to be explained. The rcm model can ‘explain’ structures or ‘macroprocesses’ only in terms of individual motivations whose very presence must be deduced from the structures themselves.
There is no way of getting from individual choice to historical processes without inserting all these ‘structural’ factors, either by reading them into the individual psyche or, in a ‘weaker’ methodological individualism, as part of the data on the basis of which the individual’s choices are made. If the model serves any useful purpose at all, it can only be in the presentation, but not in the construction, of an explanation.
The Game of Class Relations
But let us accept, for a moment, the vast array of social relations and structures that are already secreted within the ‘assets’ and ‘endowments’ with which individuals enter the ‘game’ of class. What of the ‘game’ itself? At this point, the individual, however richly endowed with historical and structural presuppositions, must enter into relations with other individuals and make strategic choices. This is where unequal individual endowments and ‘relations’ of relative advantage need to be supplemented by ‘rational choice’. Class position, according to Roemer, has to do with whether individuals work for themselves, work for others, or employ the labour of others. These positions are described as ‘chosen’ by people as they adopt ‘optimizing’ strategies which, given their respective endowments, may lead them to work for themselves, work for others, or employ the labour of others, or some combination of these options.
The notion that classes are ‘chosen’ immediately raises questions about the rcm model of choice. However much rcmists may appreciate that individual choices are never made in a social vacuum, there is little room in the model itself for the social construction of choice, for the many ways in which individual choices are structured not only by the social conditions of self-preservation and self-reproduction, by the determinate range of viable options made available within any system of social relations, but also by the complex mechanisms which operate to reproduce the system itself, including its cultural and ideological supports and their effects in the shaping of preferences. And the rcm model must begin anew with every isolated individual. It cannot accommodate the simple proposition that choices available to any single individual, or to a limited number, may not be available to all, even with the relevant preferences and the necessary ‘assets’, as the ‘choice’ of any particular ‘optimizing strategy’ by some individuals may make it less readily available to others or less economically viable in a competitive market. The reproduction of social relations cannot be explained within the strict requirements of this model except as the product of individual preferences; and even this individualistic focus is skewed by an exceptionally limited and crude psychology, which must assume a strong, stable and unchanging ego (not to mention perfect knowledge), influenced neither by its cultural and ideological environment nor by its own less conscious impulses.
But let us, for the sake of argument, take for granted the complex mechanisms of social reproduction and remain on the narrow terrain of ‘rational choice’. What are the implications of the proposition that classes are constituted by choices in pursuit of ‘optimizing’ strategies? The consequences are, in the first instance, obscured by Roemer’s emphasis on the necessity, even the automaticity, of the ‘choices’ which follow from the original unequal distribution of assets. It is even possible that for him the stage of ‘rational choice’ is an afterthought, into which he has been forced by his non-relational conception of exploitation. Some of his more enthusiastic disciples, however, have eagerly followed the path of ‘rational choice’ to its logical conclusions, dispensing with Roemer’s assumption of automaticity and thereby starkly exposing the major weaknesses in the rcm concept of ‘choice’.
A notable example is Adam Przeworski, whose work is significant particularly because he, more explicitly and systematically than other rcm notables, has used this approach to theorize a political stance. Przeworski proposes to counter the conventional Marxist notion which, as he describes it, takes ‘class positions as a given from which to begin the analysis’, and to replace it with a view which acknowledges that ‘individuals face choices, and one choice might be to become a worker and another choice might be to cooperate with other workers.’  Adam Przeworski, Capitalism and Social Democracy, Cambridge 1986, p. 97. The object of this is to understand the connection between social structure and individual behaviour, especially the connection between class conflict and politics, avoiding the assumption that certain class positions will automatically produce certain behaviours or political commitments.
It is undoubtedly wise not to assume that class positions will automatically produce certain behaviours, notably certain political commitments and actions. But there is more at stake here than a sensible injunction against simple-minded determinism. Przeworski is not content to argue that there is no necessary connection between class positions and political preferences. Instead, he asks us to treat class itself as an object of ‘choice’, analogous to other ‘optimizing’ choices, such as political preferences. Classes are ‘contingent’, dependent upon a ‘structure of choices’, not simply a reflex of production relations alone.  Przeworski, p. 96. In fact, entering a class—e.g. becoming a worker—is treated here as analytically equivalent to joining a trade union organization or a socialist party; both constitute ‘choices’—and any qualitative difference between the two kinds of ‘choice’ is elided.
The example constructed by Przeworski to illustrate his point reveals the presuppositions and consequences of the proposition that people choose classes as they select other ‘optimizing’ strategies. The exemplary worker is a certain Mrs Jones who for reasons of her own sells her labour power for a wage. Conventional Marxism (which, Przeworski argues, takes as given Mrs Jones’s class position as a ‘ready-made’ worker and deduces from it how she should behave) cannot, he says, answer the critical question: why is Mrs Jones a worker? We cannot assume that she has no choice. ‘After all, she owns some land, which she can perhaps sell; she is married to a machinist, who can perhaps work overtime; and she has, or will have, an accountant son, who might help her set up a resale shop.’  Przeworski, p. 97. She chooses to become a worker because, given her objectives and resources, this is her optimizing option.
To begin with, the necessity of explaining why the exceptional landowning Mrs Jones is a worker is not at all clear, nor is the explanatory power of an answer to this question. It might tell us something about Mrs Jones; but what would it tell us about other workers or about the operation of capitalism—or, indeed, about the process of ‘choosing’ classes? Could a landless Mr Smith be said to ‘choose’ his class position in the same sense as the lucky Mrs Jones? For that matter, what will we learn from this example even about the most mundanely practical conditions of class organization or politics? Of course we should not assume that Mrs Jones’s political preferences will be unaffected by the fact that she is a woman, white and Catholic, as well as a worker. But why does this oblige us to obscure the difference between ‘choosing’ to enter a class and joining a political party? What can we learn from such procedures that would not be far more misleading than any conclusions we might draw from the traditional Marxist assumption that the logic of capitalism compels people, separated from the means of production, to sell their labour-power in order to gain access to the means of subsistence? Is it really more informative to proceed as if people entered classes in order to earn pin-money?
What is clear is that Przeworski has loaded the dice, and he has done so because his model of class formation requires it. He seems aware of how absurd this example is (‘I do not cherish being an object of ridicule’  Przeworski, p. 97.), but he is apparently unconcerned about the extent to which the whole model depends upon the analytic centrality of exceptional cases. Possibilities which could conceivably be available to any one individual, no matter how exceptional, are here given the analytic weight of choices available to all individuals in order to characterize class formation as the result of ‘optimizing’ chocies. It is not only the fortunate Mrs Jones who ‘chooses’ her class position. The rcm model of class formation in general is constructed on the assumptions of this privileged condition.
Przeworski here reveals what Roemer conceals. With his Mrs Jones, he may have taken the rcm model to lengths unforeseen by Roemer, but he has not fundamentally traduced his mentor’s theory of class. It is in itself significant that there are two stages in Roemer’s theory of class—first the assets, then the rational choice—and that class relations belong to the second stage: people find themselves with assets (of mysterious origin), and then they choose to enter a class relation—i.e. to be exploited. This formulation cannot be dismissed as simply a rhetorical or heuristic device; for however much Roemer may recognize the necessity of the ‘choice’, the conceptualization of class in this way has substantial consequences. A great deal follows from the premise that class relations are secondary and contingent, entered into—or not—by choice. Above all, in order to permit a characterization of class relations in terms which do not endanger the model of choice and a free exchange between individuals with different endowments, there must be a redefinition of what is at stake in the relation between classes.
According to the rcm model, the operative principle in class relations is relative advantage—not whether one party is compelled by coercion to transfer surplus labour to another, or obliged to do so in order to gain access to the means of survival and reproduction, but rather whether and to what extent each party can ‘optimize’ by entering the relationship. Even Roemer, who is most emphatic about the necessity of these choices, cannot avoid this voluntaristic language. The ‘rational choice’ model of class formation requires that the relevant issues be presented as having to do with ‘optimization’ or relative advantage rather than with compulsion or the necessities of existence itself—and herein lies perhaps the most substantial modification of ‘classical’ Marxism, as well as a radical weakening of its explanatory power.
For ‘classical’ Marxism, compulsion is the essence of exploitative relations. In cases where direct producers—like feudal peasants—remain in possession of the means of production, the transfer of surplus is determined by direct coercion, by virtue of the appropriator’s superior force. In capitalism, the compulsion is of a different kind. The producer’s obligation to forfeit surplus is a pre-condition for access to the means of production, the means of sustaining human life itself. What compels direct producers to produce more than they will themselves consume, and to transfer the surplus to someone else, is the ‘economic’ necessity which makes their own subsistence inseparable from that transfer of surplus labour. Thus, wage-labourers in capitalism, lacking the means to carry on their own labour, only acquire them by entering into an exploitative relation with capital. This need not, of course, mean that those who are obliged to transfer surplus labour will get only the barest necessities; it simply means that the transfer is the necessary condition for their access to the means of survival and reproduction—and whatever they can acquire above and beyond that with those means. Such relations can be shown to exist even in the absence of any means of quantifying a ‘surplus’ or measuring the relative gains of producers and appropriators. We need only acknowledge that the producer’s reproduction has among its necessary conditions a relation to an appropriator who claims some part of his/her labour or product.
In defining exploitation, then, the limiting case for Marxism has to be one in which the relations between appropriator and producer are determined by the latter’s compulsion to transfer surplus labour, either because of subjection to a superior force or in order to secure access to the conditions of existence. The limiting case for rcm cannot be of this kind without making complete nonsense of the element of ‘choice’. Roemer himself sets up his model in the way he does, not in order to deny the compulsions which constitute exploitative relations, but rather to demonstrate that given unequal distribution of assets there can be exploitation even when everyone owns the means of production to guarantee their subsistence and even without a direct exchange of labour. But the model thereby acquires imperatives of its own—and one of its demands is the assumption that people are free not to enter the relationship at all.
The example cited by Carling in his explication of Roemer’s model is instructive. In this example, both Man Crusoe and Woman Friday begin with direct access to the means of securing material existence, though Friday possesses only a labour-intensive technology, while Crusoe enjoys a superior capital-intensive technology. Given their different ‘endowments’, they enter into a relationship, each one seeking to guarantee subsistence with the least effort. Although their assets are unequal, there is no suggestion that c has the power to coerce f into forfeiting surplus labour or to deprive her of the conditions of existence. The issue between them is simply the extent to whichf’s labours will be to the relative advantage of c, in the sense that f (though better off than before entering into a relationship with c) does not get the full benefit of her labours, part of which accrues to c. The ‘equality’ between c and f—their ‘equality as rational actors’—which the model demands, requires minimally that both (though not equally endowed) have direct and sufficient access to the means of subsistence. What is at stake is never more than ‘optimizing’ or ‘maximizing’. Carling then proceeds to characterize the power relations between c and f in terms of the ‘maximum damage’ or ‘ultimate sanctions’ each party can bring to bear on the other (p. 44). Here, he truly gives the game away. The ultimate penalty is measured simply in days of work: the number of days’ work over and above f’s ‘best option’ which c can impose on f by withdrawing from the ‘game’, and the number of days’ work which f’s withdrawal would subtract from c’s ‘best option’.
The model thus begins by—indeed depends upon—removing both forms of compulsion which have historically operated in class exploitation: the ‘economic’ necessities of capitalism and the ‘extra-economic’ coercions of pre-capitalist formations. Combining the best of both worlds, Carling offers us two ‘free and equal’ individuals, neither one subject to the superior force of the other, yet both in possession of the means of production. The transfer of surplus is not determined by direct coercion, nor is it a condition of access to the means of selfreproduction. The assumptions of this benign—and completely imaginary—condition are then transposed to real capitalism, where one of the parties begins with no access to the means of production and has only labour-power to sell. The issue between worker and capitalist is still presented on the principles of Carling’s limiting case, as if the issue were still just ‘optimization’. Even survival itself is effectively treated as simply a specific instance of ‘optimization’ only quantitatively different from any other good.
Lowering the Stakes
rcmists may argue that the choice to live or die is not the analytically relevant one in questions of class. Sometimes this means simply that, at least in principle, for any given individual (like Przeworski’s Mrs Jones) there may be other options than selling labour-power to a capitalist—options such as begging, street-vending, busking, living on welfare, relying on family support, or even leaving the working class to become self-employed. This is not a serious objection to regarding survival as the ultimate stake in capitalist class relations. From the point of view of explaining social and historical patterns, it makes a critical difference how widely available such options are; and a society in which, say, street-vending were as available an option as wagelabour would be structurally very different from one in which it is only an exceptional possibility. The other argument for weakening the compulsion involved in class relations is to say that selling labourpower is not the only option but the only acceptable one (perhaps coupled with Elster’s observation that a choice made out of necessity is still a choice  Elster, p. 13.). It is probably safe to say that all rcmists, and not just Roemer, recognize what is finally at stake. The important point, however, is that their model so absolutely requires that the stakes be relatively low and non-coercive that they must give a preponderant analytic weight to exceptional contingencies and marginal possibilities.
But even if we grant the premise that there are in principle other ways to survive, there still remains a critical difference between choosing ways of surviving and simply ‘optimizing’, in the sense of choosing to be ‘better off’. The rcm model depends on blurring this distinction too. Finally, the model requires us to ignore the ways in which even the means of escaping the necessity of transferring surplus labour are themselves determined by the dominant conditions of acquiring access to the means of production in any given class regime. So, for example, the welfare system and unemployment benefits in a capitalist system are determined by the logic of capitalist exploitation and its roots in the complete separation of the worker from the means of production. It is precisely because the logic of the capital relation for the worker means that s/he must sell his/her labour power or starve, that institutional means of preventing starvation had to be created—but only to the extent that they did not completely undermine the logic of the capital relation itself (for instance, by making such options too widely available).
There may be certain situations in which the stress on choice rather than on compulsion in relations of exploitation has rhetorical advantages; but on the whole, the costs of this focus far exceed the benefits. The effect is to make this theory of class finally indistinguishable from conventional stratification theories (common to both bourgeois social science and liberal ideology), in which the location of class divisions is more or less arbitrary because there is no qualitative break in the distributional continuum and no focal point of class antagonisms or conflicts of interest. In fact, in the final analysis Roemer (and indeed, Erik Olin Wright in his elaboration of Roemer’s theory of class) can give no consistent account of class which goes beyond identifying income differentials as the essential criterion.  For a discussion of this point, see Peter Meiksins, in Wright et al., The Debate on Classes. If the principal criterion of exploitation is an indirect relation of relative advantage, and especially when the issue for both parties in such ‘relations of comparison’ is simply ‘optimization’—in contrast to the Marxist theory of class exploitation which emphasizes the coercive relation between those who are compelled to transfer surplus and those who appropriate it—the unequal distribution of ‘assets’ can have no significance apart from its merely distributional consequences, its effects in producing inequalities of income.
Elster: A ‘One-Sided’ View?
Elster here departs in some significant ways from Roemer. He is aware of many of the difficulties in Roemer’s non-relational theory of exploitation and its extension to class, noting in particular its inadequacy in causal explanation (for example, in accounting for social conflict  Elster, pp. 203, 335–42.), which requires, among other things, a more systematic acknowledgement of power relations. The same strictures would apply to Roemer’s theory as those Elster levels against ‘stratification’ theories which focus on ‘relations of comparison’ as distinct from ‘relations of interaction’. Yet Elster’s ‘relations of interaction’ are themselves constricted, and limited in their explanatory value, by a similar view of what is at stake in the relation between classes, a view which is demanded by the imperatives of the rational choice model.
In his own account of rational choice models, and ‘intentional’ models in general, Elster makes reference to some passages from Marx in order to demonstrate that there are times when Marx himself adopts intentional explanations, though inconsistently and in contradiction to many of his basic assumptions. Elster’s interpretation of these passages is revealing. It is not the object here to deny Marx’s emphasis on human agency, choice and purpose, but rather to demonstrate how Elster’s interpretation of Marx is distorted by his own understanding of the issues at stake in the relation between classes. Elster cites the following passage from the Grundrisse as an example of Marx’s emphasis on choice:
[The worker] is neither bound to particular objects, nor to a particular manner of satisfaction. The sphere of his consumption is not qualitatively restricted, only quantitatively. This distinguishes him from the slave, serf etc. Consumption certainly reacts on production itself, but this reaction concerns the worker in his exchange as little as it does any other seller of a commodity . . . [The] relative restriction on the sphere of the workers’ consumption (which is only quantitative, not qualitative, or rather only qualitative as posited through the quantitative) gives them as consumers . . . an entirely different importance as agents of production from that which they possessed e.g. in Antiquity or in the Middle Ages, or now possess in Asia.  Elster, p. 12, Marx, Grundrisse, Harmondsworth 1973, p. 283.
Elster’s purpose is to demonstrate a contradiction (he is generally more interested in seeking out inconsistencies in Marx’s theory than in working out its fruitful insights) between this emphasis on the worker’s consumer choice and passages in which ‘Marx comes close to suggesting that the consumption of the worker is uniquely determined by his need to reproduce his labour-power.’  Elster, p. II. Marx, argues Elster, ‘had strong theoretical reasons for wanting to keep workers’ consumption fixed, since otherwise the labour value of goods might depend on preferences.’
Without engaging in a dispute about the labour theory of value, we should note the confusion of issues in Elster’s interpretation. In the ‘non-choice’ passages, Marx is seeking to explain the conditions in which the worker under capitalism gains access to the means of subsistence through the exchange of labour-power for a wage. In the Grundrisse passage cited by Elster as an example of Marx in his rcm mode, he is analysing the exchange between labour and capital first in a ‘one-sided’ way, that is, simply in the ‘sphere of circulation’. Marx makes it clear, in words elided from Elster’s quotation, that he is for the moment examining the capitalist relation incompletely, ‘as regards mere circulation’. At this level of analysis, the freedom of choice may be relevant, as is the ‘equality’ between capital and labour as parties to the exchange. Marx is here also alluding to a unique relation between consumption and production which characterizes capitalism. But he goes on to emphasize that the worker’s equality and liberty as a party to the exchange (as well as his/her freedom as a consumer) has as its presupposition ‘an economically different relation—outside that of exchange’,  Grundrisse, p. 284. a relation which is indeed masked by the exchange. It may appear to the worker, and ‘to a certain degree on the other side’, that the object of the exchange is (on both sides) to obtain ‘exchange value’ or wealth (with the possibility of such an illusion being specific to capitalism). But, Marx continues:
what is essential is that the purpose of the exchange for him [the worker] is the satisfaction of his need. The object of his exchange is a direct object of need, not exchange value as such. He does obtain money, it is true, but only in its role as coin; i.e. only as a self-suspending and vanishing mediation. What he obtains from the exchange is therefore not exchange value, not wealth, but a means of subsistence, objects for the preservation of his life, the satisfaction of his needs in general, physical, social etc. It is a specific equivalent in means of subsistence, in objectified labour, measured by the cost of production of his labour.
The last sentence, of course, suggests that Marx is even here adhering to precisely the view which this passage, according to Elster, is supposed to contradict. This should have been enough to put Elster on his guard against facile accusations of inconsistency. But the critical point is that, while at one level of analysis—that which relates to the one-sided sphere of circulation—the worker’s consumer choice may be relevant, at the other—in the account of the fundamental relation between capital and labour which is ‘presupposed’ by the exchange between them—what is essential is the necessity of the exchange as a means for the worker to secure the conditions of survival and reproduction. The fact that workers are not constrained by their relation to the means of production to choose any specific food, for example, is not what determines their position as exploited producers. The important factor here is that, since people are not in any meaningful sense free to choose not to eat, their situation is determined by the conditions in which they may gain access to food at all.
It is significant that the issue for Elster is an alleged contradiction in Marx between two different accounts of the worker as consumer. He has a tendency in general to depict class struggle as the ‘maximization’ of ‘consumer bundles’, and shows a fondness for Weber’s characterization of class systems in terms of different kinds of ‘market’ relations. Among the theories of development he attributes to Marx, he seems happiest with a periodization (of his own construction) in which the ‘dynamic element’ is trade.  Elster, pp. 310–17, 180 ff.Elster, in short, generally stops at the ‘one-sided’ analysis which remains in the ‘sphere of circulation’, indeed seems wholeheartedly to adopt the illusion fostered by capitalism concerning what is at issue in the capital relation, without unmasking the other relation which is the ‘presupposition’ of the exchange between capital and labour.
Elster does, after all, have what he would call ‘strong theoretical reasons’ for characterizing the relation between capital and labour in this one-sided way, since it is only as an exchange ‘as regards mere circulation’ that the rational choice model makes any sense at all. The model is not much use—and the notion of ‘choice’ has not much meaning—in explaining the ‘presuppositions’ of the exchange between capital and labour. Or, to be more precise, the model can deal with those presuppositions only by transposing to them assumptions derived from the ‘sphere of circulation’.
In fact, the rcm model in general seems to depend upon a resolute adherence to the ‘sphere of circulation’, in the manner of bourgeois ideology. More than that, the premises on which the whole model is based, even as it is applied to non-capitalist societies, represent a generalization of assumptions specific to capitalism—the assumptions of ‘freedom’, ‘equality’, and market-rationality—and to capitalism only as viewed in a ‘one-sided’ way.
Costs and Benefits
If the object of Rational Choice Marxism is to improve on classical Marxist explanation by breaking down ‘macrostructures’ into ‘microfoundations’, it seems not to have got us very far. At best, this procedure yields very modest results. Just how modest the pay-off really is has been persuasively illustrated by Levine, Sober and Wright in their discussion of Elster’s attempt to subject the problem of class formation to the rigours of his methodological individualism. The results produced by their attempt to rescue some explanatory power for Elster’s methodological individualism, while in some respects uncontentious and sensible, are on the whole predictable and uninteresting. ‘Class formation’ here has nothing to do with the historical processes which produce class relations. Instead, it refers to the process of organization ‘by which classes are constituted as collective actors in class struggles’  Levine, Sober and Wright, p. 80.–and even this in a very limited sense. At this level of analysis, the problems of ‘class formation’ are simply the problems faced by organizations in the attempt to mobilize support within their potential class constituencies. We learn, for example, that, since individuals are more likely to engage in action if they have some assurance that they will not be left holding the bag, organization and leadership are needed to provide an ‘indirect communication network’ which will convince them ‘that they will not be “suckers” in collective struggles.’  Levine, Sober and Wright, p. 82.This is, to be sure, a useful axiom of political organization, but it is scarcely a startling revelation (or one that requires rcm theory); and if this is the depth of insight offered by rcm, it hardly seems worth the trouble.
Besides, there are costs to offset even these modest benefits. The adjustment of the stakes in the game of class to meet the needs of the rcm model illustrates the extent to which the (purely ‘heuristic’?) form of the game-theoretic method has been allowed to dictate the substance of rcm theory. This may mean, among other things, that there is a closer connection between the substantive theses of Marxism and its methods or ‘tools’ than rcm allows, making it rather more difficult to carry out the rcm proponents’ often stated project of separating the two in order to defend the substance of Marxism by conventional analytic methods or ‘the standard tools of microeconomic analysis’.  Roemer, Free to Lose, p. 172. There is precious little left of historical materialism by the time rcm has finished begging all its questions and trivializing the stakes in the game. By returning to a distributional model of capitalism and severing the connection between the various ‘moments’ of the capitalist process, the rcm theory of exploitation has overturned every one of the central principles which Marx laboured so long to establish in his critique of political economy. If this reconceptualization had produced a paradigm with a superior explanatory power, such departures from classical Marxism could only be welcomed. As it is, we seem to have moved backward to a pre-Marxist understanding instead of forward beyond Marxism.
And one more curious result: this approach, for which Alan Carling has claimed the distinction of solving the old conundrum of structure and subject by ‘the reinstatement of societies as sets of relationships among individuals’  Carling I, p. 30. (what, then, are social relations for ‘classical’ Marxists like Marx himself?), has instead produced exactly the opposite effect. Laden with ‘endowments’ in the form of a whole system of social relations and structural necessities, the ‘individual’ turns out to be an embodied structure; and when the moment of ‘choice’ arrives, there is nothing important left to do. In the end, the principal function of ‘choice’ in the model is not to reinstate the subject but to render class contingent and ineffectual as a historical and political force.
3. Is There an rcm Theory of History?
We need to be reminded why Marxism ascribes a determinative primacy to class struggle. It is not because class is the only form of oppression or even the most frequent, consistent, or violent source of social conflict, but rather because its terrain is the social organization of production which creates the material conditions of existence itself. The first principle of historical materialism is not class or class struggle, but the organization of material life and social reproduction. Class enters the picture when access to the conditions of existence and to the means of appropriation are organized in class ways, that is, when some people are systematically compelled by differential access to the means of production or appropriation to transfer surplus labour to others.
It is no doubt possible to identify transfers of surplus labour which are not determined by such coercive imperatives (e.g. gifts, the fulfilment of kinship obligations), but these are not the kind to which the concept of class specifically refers. It is also important to acknowledge that class may not always entail direct relationships, in the sense of face-to-face confrontations, between exploiter and exploited, and that in the absence of such confrontations, class relations may not generate conflict as readily as other, more direct non-class antagonisms may do.  See Elster, pp. 338 ff. But class conflict has a particular historical resonance because it implicates the social organization of production, the very basis of material existence. Class struggle has a distinctive potential as a transformative force because, whatever the immediate motivations of any particular class conflict, the terrain of struggle is strategically situated at the heart of social existence. A trivialization of the issue between classes cannot but deprive the concept of class of its explanatory power. There can be no denying that it is difficult to imagine epochal transformations generated by games in which the stakes are relative advantages in access to ‘consumer bundles’.
Given these limitations, what kinds of claims can be made for rcm in the explanation of history? It is not altogether clear how much Roemer et al. want to claim for themselves. Carling, as we have seen, appears to concede that on the question of historical transformations—from one system of exploitation, one regime of property, one distribution of endowments, one mode of production, to another—rcm theory must remain ‘parasitic’ on some other general theory of history. Roemer sometimes (though not always  In ‘Methodological Individualism and Deductive Marxism’, for example, Roemer seems to be arguing that methodological individualism of the type advocated by Elster is essential to the explanation of historical transformations and superior to conventional Marxism in this respect.) appears to share this modest view, generally refraining from any claim to provide a dynamic model of transformation, as distinct from a kind of ‘comparative statics’—even a theory of capitalist accumulation or crisis, let alone a theory of epochal transformations. This does not, however, prevent him from associating his own theory of exploitation and class with the theory of historical materialism as interpreted by G.A. Cohen. In general, Carling has struck more or less the right note by drawing a line between rcm’s efforts to construct a theory of exploitation and class, and its attempts to establish a connection with a theory of history borrowed from elsewhere.
In his review of Free to Lose, Carling addresses himself more specifically to Roemer’s ventures into history. One of the major reservations in this otherwise respectful review concerns Roemer’s treatment of the theory of history. The objection is that Roemer fails to resolve what appears to be a fundamental contradiction between the two principal accounts of history on which he draws: Cohen’s theory of history, and Robert Brenner’s historiography, the latter being ‘the most systematic presentation of the evidence in the case [the development of capitalism] which is the acid test for the truth of the whole theory.’  Carling II, p. 95. The relevant reference for G.A. Cohen is Marx’s Theory of History: A Defence, Oxford 1978. The principal references for Robert Brenner are ‘The Social Basis of Economic Development’, in J. Roemer, ed., Analytical Marxism, Cambridge 1985; ‘The Origins of Capitalist Development: A Critique of Neo-Smithian Marxism’, nlr 104, July/August 1977, pp. 25–92; and The Brenner Debate: Agrarian Class Structure and Economic Development in Pre-Industrial Europe, T.H. Aston and C.H.E. Philpin eds., Cambridge 1985. The problem, Carling suggests, is the inconsistency between Cohen’s attribution of primacy to the forces of production and Brenner’s to class struggle; and he questions the force of Roemer’s facile response that Brenner’s evidence concerning the emergence of capitalism in the historically specific conditions of England does not contradict Cohen’s reading of historical materialism, with its general law of technological determinism, because a difference of ‘a few centuries’ in the development of capitalism in different parts of Europe hardly matters.  Carling II, p. 95, and Roemer, Free to Lose, pp. 123–4.
The question for Carling is which of the two accounts of history is a more suitable match for Roemer’s theory of class and exploitation. Roemer, he says, seems to be offering us ‘“Cohen on history plus Roemer on class and exploitation” as standard contemporary Marxism’; but, suggests Carling, given Cohen’s use of functional explanation, as distinct from what he takes to be Brenner’s rational-choice explanation, it might be more methodologically consistent to propose ‘Brenner on history plus Roemer on class and exploitation’—even though this option would be taken ‘at the cost of depriving Marxism of the support it traditionally thought it enjoyed from the theory of history’.  Carling II, p. 95.
There are several points to be made about this judgment, even before we attempt to choose between ‘Cohen plus Roemer’ and ‘Brenner plus Roemer’. Posing the question in this way—Roemer plus what?—reaffirms the discontinuity between rcm and the theory of history. The implication is that the theory of exploitation and class, for which the rational choice model is supposed to be well adapted, cannot generate its own theory of history and may indeed be incompatible with any theory of history at all. This unhappy conclusion is based not only on a clear-sighted recognition of rcm’s undynamic character, but also on a particular view of what counts as a theory of history. Like other rcmists, not least Roemer himself, Carling has no doubt that a theory of history (as distinct from, say, a ‘special’ theory of capitalism) must be a super-general, indeed transhistorical, account which posits some universal law of historical change in a determinate direction—not even just some common mechanism (such as class struggle, the outcome of which is variable and indeterminate and which has its own historically specific rules and conditions in every particular social form), but rather a general law, a ‘deep cause’, that transcends all historical particularities. Thus Cohen has a theory while Brenner has only historiography.
An attempt to resolve the Cohen-plus-Roemer v. Brenner-plusRoemer issue may tell us as much as we need to know about rcm and its relation to the theory of history. Let us first try to reconstruct Roemer’s argument. One thing is clear: Roemer, despite some reservations, still associates himself with historical materialism as a theory of history, and for him this means (evidently without question) Cohen’s technological determinism.  Roemer, Free to Lose, p. 108. It remains to be seen whether he is wrong to think—as Carling suggests he may—that his own theory of exploitation and class can be consistently coupled with Cohen’s theory of history, and whether Carling is right to propose that Brenner would make Roemer a more suitable match.
Roemer on History
History, according to Roemer, takes the form of an evolution in property relations, in which ‘progressively fewer kinds of productive factors remain acceptable as property’.  Roemer, Free to Lose, p. 126. For example, property in persons is eliminated as slave society passes into feudalism, leaving some property rights in the labour of others and property in alienable means of production. The transition from feudalism to capitalism eliminates property rights in the labour of others while still allowing property in the alienable means of production, and so on. This ‘progressive socialization of property’ occurs for reasons ‘related to efficiency’, that is, the advancement of the forces of production. ‘The mechanism that brings about this evolution is class struggle’, but ‘the reason such an evolution occurs lies somewhere deeper: evolution occurs because the level of development of the technology outgrows the particular form of the social organization, which comes to constrain and fetter it.’  Roemer, Free to Lose, p. 6. Emphasis in the original. Given rcm’s need to lower the stakes in the game of class, it may be significant that Roemer chooses to emphasize the progressive socialization of property as the elimination of exploitative forms, in contrast to Marx’s focus on history as the progressive separation of direct producers from the conditions of their labour (this is, for example, the basic premise of the Grundrisse).
The connection between the mechanism (class struggle) and the deep cause (technological determinism) can be explained in the following way. Class struggle serves as a ‘facilitator’ in the transition from one social form to another, when the ‘dissonance’ between the level of development of productive forces and the old economic structure reaches a crisis point. So, for example, Roemer asks us to ‘imagine’ (his word) a feudal system, with lords and serfs, but one in which ‘a nascent capitalist economy is emerging alongside’ the feudal system.  Roemer, Free to Lose, p. 115.
‘Now there is an option: capitalists and feudal lords can compete for control of the working population. If the technology or forces of production that the capitalists are using enables them to pay higher real wages than serfs can earn, then there is an economic advantage to the liberation from serfdom that did not formerly exist.’ Serfs can become independent peasants, taking advantage of the trade opened up by the capitalists, or they can become artisans or proletarians in the towns. ‘The competition between feudalism and capitalism now enables class struggle against feudalism to be successful even though formerly it was not.’
We now have three levels of explanation: (1) the deep cause (technological determinism); (2) the historical process (the successive elimination of forms of exploitation or the progressive socialization of property); (3) the ‘facilitator’ (class struggle—though this only ‘facilitates’ a process that was ‘bound to come sooner or later’  Roemer, Free to Lose, p. 124.). It is not entirely clear at what level the rational choice model should be introduced. The most obvious place is in class struggle, which implies that change occurs when (if not because) people are in a position to choose the available option of the next, more progressive mode of production. At the same time, there appears to be an overarching rational choice, at the level of the deep cause, having to do with ‘the ceaseless effort of rational human beings to alleviate their conditions of scarcity’  Roemer, Free to Lose, p. 123.—though they do not actually choose the next available economic structure because it is conducive to technological progress. In either case, the necessary link between the rational-choice model and the theory of history is the presupposition that there is a direct correspondence between the self-interested actions of individual rational actors and the requirements of technical progress and economic growth.
This three-layered structure raises rather more questions than it answers, not least about the connections among its three levels. Are rational individuals, insofar as they are the makers of history (but are they?), motivated by the desire to alleviate scarcity through technological improvement or by the wish to escape exploitation—or neither? Is class struggle necessary or not; and if not, what is the mechanism of historical change? Or does the deep cause somehow make mechanisms and facilitators redundant, since change is somehow ‘bound to come sooner or later’ anyway, behind the backs of rational individuals? And where, in any case, is the struggle in class struggle? We have lords and capitalists competing to give more attractive terms to producers, to serfs who might want to become proletarians; and we have serfs escaping from lords—apparently without constraints, and willingly giving up their rights of possession—as soon as a more attractive option comes along; but struggle. . . .? What, indeed, is the ‘economic advantage’ which would impel serfs to prefer a wage which, by some sophisticated statistical measure, was higher than their ‘earnings’ as serfs, at the cost of losing their rights of possession, giving up the land which provides them with full and direct access to the means of subsistence in exchange for the uncertainties of the proletarian condition? For that matter, even if serfs choose this option, how do they manage to achieve it? If the lords’ property ‘rights’ in the labour of others have anything to do with the ‘control’—i.e. the power—that they exercise over the serfs, what is the nature of that ‘control’, and how is it that when the critical moment of transition comes, serfs can simply choose to escape the lord’s ‘control’ just because a more eligible option has presented itself? Has feudalism no self-sustaining logic and resources of its own to resist this easy transition?
All this is quite apart from the fact that the whole edifice is constructed without benefit of evidence. Roemer has chosen his words carefully when he asks us to ‘imagine’. We can hardly do anything else. (‘Imagine’ and ‘suppose’ are the basic vocabulary of this gametheoretic discourse.) We are not (or not always) being asked to believe that this is how things actually happened, or even that it was historically possible for them to happen in this way, only that it is logically conceivable that they did (though it is never made clear why we should be interested in such imaginary logical possibilities). Indeed, it is very unlikely that Roemer himself believes his own imaginary account of the transition to feudalism; and it is a measure of the price exacted by his game-theoretic model that it obliges him to set aside everything he undoubtedly knows about the power relations between lords and serfs, the dispossession of small producers and the concentration of landlordly property which were the conditions of the transition; everything, that is, about coercion, compulsion, imperatives, or indeed about the social relations of exploitation. This process of transition which Roemer asks us to ‘imagine’ evidently has little to do with history, and it would serve no purpose to counter this imaginary story with evidence. History is, apparently, a subject about which we can say anything we like.
Begging the Question
There is one thing that Roemer asks us to ‘imagine’ which is inescapably critical to his argument, an assumption upon which the whole shaky edifice rests and which goes to the very heart of rcm and its relation to history. We must accept that capitalism already exists as an ‘option’, that a ‘nascent capitalist economy is emerging alongside’ the feudal system. We must also never ask how this came to be so, though for Roemer this evidently presents no problem:
According to historical materialism, feudal, capitalist, and socialist exploitation all exist under feudalism. At some point feudal relations become a fetter on the development of the productive forces, and they are eliminated by the bourgeois revolution. . . . Historical materialism, in summary, claims that history progesses by the successive elimination of forms of exploitation which are socially unnecessary in the dynamic sense.  Roemer, General Theory, pp. 270–71.
He goes on to characterize this interpretation as ‘a translation of the technological determinist aspect of historical materialist theory into the language of the theory of exploitation’. According to this interpretation of historical materialism, all successive forms of exploitation are already contained in the preceding ones (his account of the progress from slave society to capitalism suggests that this retrospective analysis goes back beyond feudalism), so that all forms of exploitation which have emerged in the course of history have apparently been present since the beginning; and history proceeds by the process of elimination.
A similar conflation has, of course, already occurred at the elemental level of Roemer’s rational choice model, in the very foundations of his general theory of exploitation, where every mode of exploitation is defined in terms of its alternative, the successor mode which is already (or rather, still) present in the existing one. Remember, for example, the ‘test’ for feudal exploitation in terms of its ‘hypothetically feasible alternative’: ‘to calculate whether a person would be better off under the distribution where no feudal property exists but capitalist property still exists.’  Roemer, Free to Lose, p. 136. In Roemer’s account of history, the ‘hypothetical’ alternative turns out to be an actually existing one, indeed the very next historical stage which happens to be already present in embryo.
It has been a favourite ploy of theorists who have trouble with process to beg the question of history by assuming that all historical stages—and especially capitalism—have in effect existed, at least as recessive traits, since the beginning. The perennial quality of capitalism has been a staple of bourgeois ideology, and also a common assumption of influential historians like Henri Pirenne. The usual approach in those Marxisms that go in for this kind of thing—the most recent and well-developed example is Althusserian structuralism—is to conjure up images of aspirant modes of production lurking in the interstices of previous ones, waiting only for the opportunity to establish their ‘dominance’ when certain obstacles are removed. The Althusserian concept of ‘social formation’, in which any and all modes of production can coexist without any need to explain their emergence, has done yeoman service in this regard. But Roemer has truly perfected this strategy, with his theory of elimination (it seems so much easier somehow to account for the demise of what already exists than to explain its coming into being). There can be no doubt that this kind of conceptual conjuring has far too often served in place of a Marxist theory of history; and, no doubt, Marx’s more formulaic aphorisms about the stages of history and successive modes of production could be read as invitations to evade the issue of historical processes in this way. It was Marx, after all, who first spoke of ‘fetters’ and ‘interstices’. But there is much more in Marx which demands that we look for the key to historical change in the dynamic logic of existing social relations without assuming the very thing that needs to be explained.
Brenner plus Roemer?
This is the point at which we can best judge the compatibility of Roemer’s rcm with Robert Brenner’s approach to history, for Brenner’s primary purpose has been precisely to break the prevailing habit of begging the central historical question, the practice of assuming the existence of the very thing whose emergence needs to be explained. He distinguishes between two kinds of historical theories in Marx’s own work, the first still heavily reliant on the mechanical materialism and economic determinism of the eighteenth-century Enlightenment, the second emerging out of Marx’s mature critique of classical political economy. The first is characterized precisely by its begging of the question, invoking the self-development of productive forces via the division of labour which evolves in response to expanding markets, a ‘nascent’ capitalism in the womb of feudal society.
The paradoxical character of this theory is thus immediately evident: . . . there really is no transition to accomplish: since the model starts with bourgeois society in the towns, foresees its evolution as taking place via bourgeois mechanisms [i.e. change and competition leading to the adoption of the most advanced techniques and to concomitant changes in the social organization of production. emw], and has feudalism transcend itself in consequence of its exposure to trade, the problem of how one type of society is transformed into another is simply assumed away and never posed.  Robert Brenner, ‘Bourgeois Revolution and Transition to Capitalism’, in The First Modern Society, A.L. Beier et al., eds., Cambridge 1989, p. 280.
Later, Marx was to pose the question very differently. He substantially revised his views on property relations in general and pre-capitalist property relations in particular: ‘In the Grundrisse and Capital, Marx defines property relations as, in the first instance, the relationships of the direct producers to the means of production and to one another which allow them to reproduce themselves as they were. By this account, what distinguishes pre-capitalist property relations . . . is that they provided the direct producers with the full means of reproduction.’ The condition for maintaining this possession was the peasant community, and its consequence was that the lords required ‘extraeconomic’ means of taking a surplus, which in turn demanded the reproduction of their own communities. The structure of these property relations was thus reproduced ‘by communities of rulers and cultivators which made possible the economic reproduction of their individual members.’  Brenner, ‘Bourgeois Revolution’, p. 287. Given these property relations, reproduced by communities of rulers and direct producers in conflict, the individual lords and individual peasants adopted the economic strategies that would best maintain and improve their situation—what Brenner calls their rules of reproduction. The aggregate result of these strategies was the characteristic feudal pattern of development. The transition to a new society with new developmental patterns thus entailed not simply a shift from one mode of production to another alternative mode, but a transformation of existing property relations, from feudal rules of reproduction to new, capitalist rules. Under these, the separation of direct producers from the means of production, and the end of ‘extra-economic’ modes of extraction would leave both appropriators and producers subject to competition, and able to move in response to the requirements of profitability under competitive pressures. This inevitably raised a new and different question about the transition from feudalism to capitalism: ‘it was the problem of accounting for the transformation of pre-capitalist property relations into capitalist property relations via the action of pre-capitalist society itself.’  Brenner, ‘Bourgeois Revolution’, p. 293.
This is the challenge which Brenner has taken up: to offer an explanation of the transition to capitalism which relies entirely on the dynamics of feudal relations, and its conditions of reproduction, without reading capitalism back into its predecessor or presenting it as an available option.  Brenner’s contribution to the collection on Analytical Marxism edited by Roemer, which presents his account of the transition in a schematic form, argues that the conventional explanation of the development of capitalism, based largely on Adam Smith, assumes precisely the extraordinary phenomenon that needs to be explained; that property relations must be understood as ‘relations of reproduction’; that precapitalist economies have their own logic and ‘solidity’, which are in effect denied by the conventional view; that capitalist development is a more historically limited and specific occurrence than is allowed by theories which attribute it to some universal law of technical progress; and that the history of the transition cannot be explained by assuming that there is a necessary correspondence between the self-interested actions of individual actors and the requirements of economic growth. In these respects, his argument runs directly counter to Roemer’s basic assumptions, and to Cohen’s technological determinism. It is worth adding that his schematic argument in this volume, in contrast to the contributions of Rational Choice Marxists, is grounded in historical work and is based on the premise that the work of historical explanation needed to be done in advance of the analytic presentation. This project also requires an acknowledgement that pre-capitalist property relations have a logic and tenacity of their own, which cannot be conjured away by the convenient assumption that people are driven by an urge to take the next available (capitalist) option, an urge that existing structures cannot resist. This is something which Roemer’s model is systematically unable to take into account. In this respect, Roemer’s model reproduces those of Adam Smith, the early Marx, Henri Pirenne, and Paul Sweezy, which Brenner has consistently criticized. They too, he argues, tend to assume the existence of capitalism in order to explain its coming into being. Roemer would be subject to the same criticisms levelled by Brenner against what he calls ‘Smithian Marxists’, because, on the rational choice reasoning to which Roemer is committed, he simply cannot get from feudalism to capitalism without already assuming capitalist structures and capitalist motivations. The relations of capitalism and the associated compulsions of capital accumulation, the specific logic of capitalism and its systemic imperatives, can in no way be deduced simply from the unequal distribution of ‘alienable assets’ which for Roemer constitutes the kernel of capitalism already existing in feudalism. (Here, the inadequacy of Roemer’s model of exploitation—defined in terms of inequality or relative advantage—becomes most starkly obvious, since whatever value it may have in explicating the injustices of exploitation, it is completely incapable of apprehending capitalism as a particular system of social production, appropriation and accumulation.) Nor can the relations and imperatives of capitalism be deduced from the mere existence of towns, on the assumption—unwarranted both logically and historically—that towns are by nature capitalist.
Brenner’s approach represents a challenge to the rcm model at every level of analysis, from the most specifically empirical to the abstractly theoretical. Empirically, Brenner’s history of capitalist development puts in question nearly every point in Roemer’s imaginary scenario. Capitalism does not, in his account, simply exist, miraculously, ‘alongside’ the feudal economy, nor is it here a product of mercantile interests in the towns competing with feudal interests in the countryside. Direct producers do not join the capitalist economy by fleeing the countryside to become artisans or proletarians. The development of capitalism is rather a process set in train by the transformation of agrarian relations themselves, in particular conditions which have little to do with some exogenous expansion of trade. Indeed, this account (like others in the famous ‘Transition debate’  Rodney Hilton, The Transition Debate, Verso, London, 1976. See also Hilton, ‘Towns in English Feudal Society’, in Class Conflict and the Crisis of Feudalism: Essays in Medieval Social History, London 1985.) begins by casting doubt on the inherent antagonism of markets and trade to the feudal order. It is not capitalism or the market as an ‘option’ or opportunity which Brenner seeks to explain, but the emergence of capitalism and the capitalist market as an imperative. His is a history of the very special conditions in which direct producers in the countryside were subjected to market imperatives, rather than the emergence of ‘options’ for direct producers, opportunities offered them by trading interests in the towns.
More generally, the facts of the case as presented by Brenner run counter to Roemer’s ‘deep cause’ (and Cohen’s technological determinism)—as Roemer himself is apparently aware—by indicating that there is no historical necessity for less productive ‘economic structures’ to be followed by more productive ones and by stressing the historical specificity of the conditions in which the process of ‘selfsustaining’ growth was first established. Roemer’s response to this challenge, readers may recall, is simply that capitalism did eventually spread to other parts of Europe, even if it first emerged in England, so that Brenner’s evidence does not contradict Cohen’s version of historical materialism, with its universal law of technological determinism. This response depends on a cavalier treatment of historical time (who cares about ‘a few centuries’?), as Carling has observed, and a similar treatment of geographical space (who cares about the world beyond Europe—or indeed beyond western Europe?). More fundamentally, it depends on treating the historically specific process of capitalist expansion, a priori, as a transhistorical law of nature.
When Roemer invokes the universality of capitalist development, at least its eventual spread to other parts of Europe, this is no reply to Brenner. It is, again, a begging of the question. The whole point of Brenner’s argument is to avoid simply taking for granted the universal development of capitalism by subsuming it under some universal law of historical change, and instead to explain how historically specific conditions produced capitalism’s unique drive and its unique capacity for expansion and universalization, from historically specific origins. The very least that can be said about the inadequacy of Roemer’s response is that the universality of capitalism can be deployed as evidence for Brenner’s account just as easily as for Cohen’s. Brenner’s has the advantage of resting on a wealth of historical evidence and of not assuming the very thing that needs to be explained.
What Is a Theory of History?
What is at stake here is not, however, only Brenner’s empirical challenge to Roemer-Cohen’s theory of history. Brenner’s explanation of capitalist development as emerging out of feudal relations themselves, instead of ‘alongside’ them, plays havoc with the very foundations of the rcm model. Historical change is generated in his account by processes within existing social relations, and not by the ‘choice’ of some other, exogenous or co-existing, ‘option’. Certainly lords and peasants make ‘rational choices’ (what serious historian would deny this?), but those choices occur within the existing relations. Indeed, they are aimed not at attaining the next, more attractive, historical stage (which the parties involved cannot, in any case, anticipate) but at the reproduction of existing conditions.
Such an account implies something more than individuals with (fixed and static) ‘resources’ or ‘endowments’ plus ‘rationality’ (with magically available options). It implies a whole historically constituted and dynamic network of structured relations—between lord and peasant, between each and his community, between class and state—always in process; and it cannot beg the question of historical origins. Brenner’s contribution to ‘analytic Marxism’ (explicitly stated in the volume of that name edited by Roemer) is to break the link which ties the gametheoretic model to the theory of history: he undermines the presupposition that there is a direct correspondence between the self-interested actions of rational economic actors and the requirements of technical progress or economic growth. If there is anything distinctive in rcm, if it is more than just an ‘analytic’ mode of presentation or even a very general acknowledgement that people make history by making choices which may have unintended consequences (if, for example, rcm is serious about constructing its game-theoretic universe out of just individual resources plus rationality) then Brenner’s history represents a challenge to its very first methodological premises.
It should already be evident that Brenner’s approach also challenges the whole notion of historical theory as a general account of universal laws which move in a pre-determined direction. Not only does he question the historical-materialist credentials of such a theory by attributing it to an undeveloped phase of Marx’s work, still uncritically bound to classical bourgeois thought; but his whole historical project testifies to a view that a theory of history subsuming the entire developmental process from classical antiquity to capitalism (let alone the whole of world history) under one universal and essentially unidirectional law of motion, would need to be so general as to be vacuous. How useful, after all, is a ‘theory’ of technological progress which claims to be equally compatible with moments of rapid technical improvement and long epochs of stagnation or ‘petrification’?  It was Marx himself who insisted that capitalism is unique in its drive to revolutionize productive forces, while other modes of production have tended to conserve existing forces, and that ‘petrification’ may have been the rule rather than the exception. See, for example, Capital I, Moscow 1971, pp. 456–7. A similar view appears even in the Communist Manifesto, which in other respects still adheres to the early, uncritical theory of history. I have criticized Cohen’s technological determinism elsewhere, in ‘The Separation of the Economic and the Political in Capitalism’, nlr 127, pp. 70–74, and in a more general discussion of technological determinism in ‘Marxism and the Course of History’, nlr 149, pp. 95–107.
To say this is not to accept Carling’s view that Brenner has empirical evidence but not a theory of history. It is instead to suggest that there may be other conceptions of theory, and indeed that the most distinctive feature of historical materialism—that which distinguishes it most radically, in form as well as substance, from conventional bourgeois theories of progress—is not its adherence to a general law of technological determinism. It is, rather, a focus (such as that which characterizes the most complete and systematic of Marx’s own works, his actual practice in the critique of political economy and the analysis of capitalism) on the specificity of every mode of production, its endogenous logic of process, its own ‘laws of motion’, its characteristic crises—to use Brenner’s formula, its own rules of reproduction.
This is not simply a matter of distinguishing, as Carling does, between a ‘general theory’ of history and a ‘special’ theory of capitalism. It is rather a different (and general) theory of history of which the theory of capitalism, with its specific laws of motion, is the prime example. While Cohen’s ‘general theory’ takes the form of retrospective or even teleological predictions, with the benefit of hindsight, with such a degree of generality that no empirical evidence could possibly falsify it, Brenner’s theory demands empirical specification which does not assume a predetermined outcome. But if the mark of a theory is the existence of ‘fixed points’ which remain constant throughout its specific applications, there are more than enough fixed points in Brenner’s account—in particular, the principle that at the foundation of every social form there are property relations whose conditions of reproduction structure social and historical processes.
Cohen plus Roemer?
Does this mean that ‘Cohen plus Roemer’ works better? There are, of course, problems here too, not the least of which is that Cohen’s account of technological determinism, with its emphasis on ‘functional explanation’, seems to run counter to all the rcm strictures against ‘methodological collectivism’. And yet, on second thoughts, the attraction which Cohen exerts upon Roemer and other rcmists may be understandable and not at all inconsistent, or at least it is entirely consistent with the irreducible contradiction at the heart of rcm.
rcm requires that, if functional explanations are invoked, a mechanism must be found to account for how the ‘functional’ effect is produced. Cohen’s functional explanation of the development of productive forces rests on a particular theory of human nature which purports to supply the necessary mechanism: given a rationality which involves a disposition to seize the means of satisfying compelling wants, and a kind and degree of intelligence which permits them to improve their situation, in conditions of scarcity, human beings will generally seek and/or adopt whatever means are available to curtail labour, which is inherently unpleasant to them (never mind evidence to the contrary, such as that provided by Brenner). It is this transhistorical rationality, and the impulse to improve productive forces which it entails, that underlies the functional argument for the primacy of productive forces.
Because of rcm’s own static model of ‘property relations’, the first requirement in an rcm-compatible theory of history is that the ‘motor’ must come from outside any given system of property or exploitation. In other words, it is in the nature of the rcm model that, in order to account for historical process, it requires a deus ex machina. Traditionally the most popular ‘external’ forces have been trade (markets enlarging and contracting, with trade routes opening and closing) and/or technical progress, both conceived as exogenous to the systems being transformed, either in the sense that they are determined by alien intrusions such as barbarian invasions, or in the sense that they operate according to some universal natural law (progress, the natural development of the human mind, or perhaps, more scientifically, demographic cycles . . . ). rcm, however, has the special requirement that the explanation must be reduced to motivations which can be attributed to the rationality of individual human beings. A particular kind of transhistorical rationality, which can serve as the common denominator among all the various systems of property and exploitation, and can supply the mechanism for moving from one to the next, in effect containing the whole of history within it, seems tailormade for the purpose. To put it another way, if historical motion can be explained only by presupposing each successive historical stage as an option or motivation already available in the preceding one, then it helps to have some exogenous cause for its availability. An rcmcompatible theory of history requires also that the ‘rational choice’ model of exploitation be left intact, and this means that production relations, which are not so much the conditions of rational choice as its product, can play no immediate part in setting off the dynamic of change. What is needed is a theory of history which, as it were, runs parallel to the analytic model without impinging upon or challenging its constitutive analytic fiction, which presents the exploitative ‘relation’ as an agreement between rational agents.
Cohen, unlike Brenner, permits rcm to maintain its static model and does not place upon it the intolerable strain of motivating historical change from within specific property relations, without the aid of exogenous forces or ‘feasible alternatives’ which appear by spontaneous generation, in the form of already existing options, motivations or ‘assets’. Thus rcm, a theory ostensibly devoted to rescuing human agency from structural determinisms, is forced to readmit a fairly oldfashioned historical determinism through the back door, while retaining the fiction of ‘rational choice’ at another, ahistorical level of analysis.
On balance, Roemer appears to be right in thinking that an alliance with Cohen is a safer bet than collaboration with Brenner. At the same time, since neither match is unproblematic, it might be more precise to say that Roemer’s theory of class and exploitation is uncongenial to any explanation of history. Anyone who believes that a vacuously general theory is little better than no theory at all in historical explanation, might be inclined to think that it is just here, in its fundamentally ahistorical character and in its hostility to historical specificity, that rcm has most in common with Cohen’s technological determinism.
Elster on Marx and History
Elster’s approach to technological determinism presents problems of its own. He is more inclined than Roemer to demand from Cohen a specific account—in methodological-individualist terms—of the mechanisms by which the long-term ‘functional’ effect of technological progress is achieved, and in this sense significantly qualifies his adherence to the ‘general theory’ and its reliance on functional explanation. He also questions the extent to which Marx himself consistently adhered to the general theory. On this point, Elster’s principal concern is to demonstrate that Marx’s theory of history is inherently inconsistent; and his argument centres on one important insight, namely that in Marx’s own accounts of historical transitions, the development of productive forces plays little role as the primary motor.
This observation is both true and significant. Although the ‘general theory’ appears in various (though not many) aphoristic formulae in Marx’s work, it is not generally deployed in his systematic attempts at historical explanation. So, for example, his most comprehensive accounts of pre-capitalist societies in the Grundrisse and of the historical transition to capitalism—especially in the section on ‘Primitive Accumulation’ in Capital—do not invoke the development of productive forces as the motivating impulse of historical change, and indeed are based on the premise that what needs to be explained is precisely the origin of capitalism’s distinctive drive to improve the forces of production. Here, in the critique of political economy which formed the core of his life’s work, Marx laid the foundation for a different kind of ‘general’ theory—the foundation on which Brenner, for instance, has constructed his own account of history. But Elster is not interested in this theoretical alternative, if he even recognizes it as such (as distinct from a ‘special’ theory of capitalism). He is interested only in demonstrating the incompatibility between Marx’s adherence to the primacy of productive forces and his analysis of capitalism, a contradiction between the general theory of history as the development of productive forces and the view that the maximization of technological development is specific to capitalism.
There is no doubt that Marx never bothered to resolve the inconsistencies between his aphorisms about the forces of production and his insistence on the specificity of capitalism. It needs to be said, however, that even in the context of the ‘general theory’ (if such it is), there is room for a specifically capitalist dynamic. The critical issue for Marx was always the specific compulsion of capitalism to revolutionize the forces of production, which differs from any more general tendency to improve productive forces that may be ascribed to history as a whole. In that sense, it was possible for him consistently to hold both the view that history displays a general tendency to improve the forces of production and the view that capitalism had a special need and capacity to revolutionize productive forces. Beginning early in his career, Marx never deviated from the view that the capitalist drive is specific and unprecedented and that, whatever progressive tendencies may be generally observable in history, the specific logic of capitalism and its specific compulsion to improve the productivity of labour by technical means are not reducible to these general tendencies. They require a specific explanation. He also made it very clear that the capitalist impulse to improve the productivity of labour is quite distinct from, and often in opposition to, any general human inclination to curtail labour. The capitalist impulse is to increase the portion of unpaid labour. He devoted much of his life’s work to explaining this specifically capitalist dynamic.
Marx’s critique of political economy, the core of his mature work, proceeds precisely by way of differentiating himself from those who take for granted and universalize the logic and dynamic of capitalism without acknowledging the historical specificity of its ‘laws’ or seeking to uncover what produces them. Marx, unlike the classical political economists, and indeed a host of other ideologues of ‘progress’ and ‘commercial society’, did not assume that the ‘progress’ embodied in modern society was simply the outcome of a drive inherent in human nature or natural law, but insisted on the specificity of the capitalist demand for productivity and the need to find an explanation for it. It was Marx’s identification of that specific dynamic which made it possible even to raise the question of the ‘transition’ to capitalism, to seek an explanation of how that dynamic was set in train, something which remained impossible as long as people assumed the very forces that needed to be explained. Marx himself never produced a systematic account of the historical process of transition, and his discussions of pre-capitalist modes of production were never more than retrospective analyses, part of a strategy to explicate the workings of capitalism and emphasize the historicity of its laws and categories. But he took the qualitative leap that was required to make an explanation of the transition possible, and he thereby established a basis for a general theory of history which would also treat other modes of production on their own specific terms.
History or Teleology?
It is especially ironic that the strategy adopted by Marx to highlight the specificity of capitalism is mistaken by Elster for a teleological account of history. This misunderstanding by itself tells us something important about Elster’s conception of what counts as a theory of history and its essential affinity to ‘Roemer-plus-Cohen’. In a particularly significant passage, Elster cites Marx’s famous aphorism that ‘Human anatomy contains a key to the anatomy of the ape’, from the Grundrisse, as a statement of his ‘teleological stance’—which is, according to Elster, ‘closely related to the propensity for functional explanation. . . .’  Elster, p. 54. Elster compounds the misunderstanding by interpreting this aphorism to apply to the ‘teleological’ relationship between communism and capitalism, as it does to the relationship between capitalist and pre-capitalist modes of production. Although capitalism can, of course, be analysed from the standpoint of socialism—that is, by identifying the potentiality within capitalism for a socialist transformation—the analogy suggested by Elster is an imperfect one. Capitalism can provide the ‘key’ to pre-capitalist society, in the sense here intended, only because it actually exists and because it has given rise to its own historically constituted categories, whose historical specificity Marx is trying to demonstrate by critically appying them to precapitalist forms. That is precisely the meaning of his critique of political economy. The point is not that capitalism is prefigured in pre-capitalist forms but on the contrary that capitalism represents a historically specific transformation. Marx adopts this paradoxical strategy precisely in order to counter ‘those economists who smudge over all historical differences and see bourgeois relations in all forms of society’  Marx, Grundrisse, p. 105.—that is, precisely against what might be called the teleological tendencies of classical political economy, for which ‘bourgeois relations’ are the natural and universal order of things, the destination of progress already present in all earlier stages of history. To the extent that Marx’s point of departure is a refusal to incorporate capitalism into the historical process which produced it, and especially with a theory of history in which every mode of production is propelled by its own distinctive laws of motion, his characteristic procedure is exactly the reverse of teleological or even functional explanation.
Although a teleological account of history represents the most serious violation of methodological individualism, rcm has a tendency to incorporate the destination of history into its very beginnings—a teleological procedure if ever there was one. On the one hand, rcm can give no account of history but must begin afresh with every social form, an already given and static ‘property relation’ represented in the person of the rational individual whose ‘endowments’ and ‘relational properties’ come from nowhere and are going nowhere; on the other hand, it is drawn to theories of history in which the end is already given in the beginning—as forms of exploitation coexisting until they successively fall away, or as a transhistorical drive for technological development, requiring only ‘elimination’ or the removal of impediments to come to the fore. In both cases, the question of historical processes is begged. Perhaps, then, it is simply the idea of process that sticks in the rcm throat, and the problem—for someone like Elster—is not that conventional Marxism is teleological but that it is historical, a distinction which may not be altogether clear in the rcm framework.
At worst, then, the rcm model as a guide to history is positively misleading. At best, it can add little to what Marxism can achieve by other less precious and circuitous means. And since the context in which ‘rational choices’ are made must always be specified first (and the model cannot help us to arrive at that specification), if the model is to be used at all in the explanation of social and historical processes, then all the real work—the historical and structural analysis—needs to be done before the model can be inserted. In such a case, the model is, again, largely rhetorical or persuasive. If that is so, we really have to ask whether the game is worth the candle. What rational being would choose rcm if the pay-off is so incommensurate with the effort?
4.The Moral Argument
We are left with the claims of rcm as a mode of moral argumentation. If it can make good Roemer’s claim that ‘the game-theoretic formulation is superior [to conventional Marxism and its ‘surplus labour’ theory—emw] in that it makes explicit what ethical presumption lies behind the Marxian theory of exploitation’, if it can thereby enhance the ideological arsenal of socialism by securing the case for its moral superiority to capitalism, then a useful purpose will no doubt have been served. And if this is rcm’s principal project, we may be inclined to conclude that its explanatory weakness is not such a serious shortcoming after all. It would certainly fill a gap in socialist thought which, among other things, has put Marxism on the defensive against the charge that the explicit disavowal of ethical considerations in the interests of ‘scientific socialism’ has disarmed it in the face of deformations such as Stalinism. This vacuum in Marxist argumentation has been more acutely felt as growing disappointment with revolutions in the East and their absence in the West have sapped an older confidence in the historical tendency toward, if not the inevitability of, socialism, which as a historical necessity could, apparently, dispense with an ethical justification.
It is, however, unlikely that such considerations have weighed very heavily with Rational Choice Marxists. They have generally shown little interest in socialist history, in its international dimension, and have been almost exclusively preoccupied with interlocutors on the academic Right. What seems to have concentrated their minds on ethical questions is the rise, since the 1970s, of an increasingly aggressive and confident intellectual Right. For the first time in recent memory, the Right has marshalled its intellectual forces, inside and outside the academy, to demonstrate that capitalism is materially and morally better than socialism, both more efficient and more free.
But if rcm is principally designed to meet the philosophical challenge from the academic-intellectual Right, it has entered the battle with (at least) one hand tied behind its back. This paradigm is singularly illdesigned to give us a sense of capitalist accumulation as a ruthless world process which inevitably generates economic catastrophes outside the privileged North, as well as regular crises within it, not to mention the threat of nuclear or ecological destruction for the world altogether. Indeed, the rcm corpus generally leaves the impression of a very limited mental universe, confined within an opulent Northern capitalism, and even there constricted by a remarkable insensitivity to the irrationalities and destructive effects of capitalist accumulation. The rcm case for socialism is generally predicated on a relatively benign capitalism—a capitalism which is particularly effective in satisfying material interests (always narrowly conceived in terms of ‘consumer bundles’). Even Roemer, who is committed to the view that capitalism is in principle anarchic and inefficient, is constrained by his overriding methodological commitment to the abstract models of conventional economics. He presents his indictment of capitalist inefficiency not by exposing the realities of its wasteful destructiveness but, indirectly, by mildly acknowledging, in very general and abstract terms, that some of the assumptions of equilibrium theory (a theory which he nevertheless regards as ‘one of the great contributions to social scientific method of the past century’) may not apply in real capitalist economies.  See, for example, Roemer, Free to Lose, p. 151. The weakness of the barrier erected by rcm against right-wing triumphalism is illustrated by John Gray’s review of Free to Lose in the Times Literary Supplement, 24 February–2 March 1989. Roemer is here castigated for adopting the worst features of neo-classical economics, its complete abstraction and detachment from the real world of market processes—though in Gray’s account, of course, the real world triumphantly vindicates capitalism and decisively demonstrates the inferiority of socialism. Nothing in Roemer’s argument equips him to withstand this onslaught. Other rcmists are evidently not prepared to concede even this much to the critique of capitalism, suggesting at best that the superior efficiency of capitalism may have to be sacrificed in the interests of some higher commitment—perhaps to democracy and collective decision-making.  Przeworski, pp. 237–8. Such a restricted analysis of capitalism not only deprives the socialist project of any historical foundation but weakens the moral case against capitalism.
rcm leaves little ground for socialist struggle against capitalism, nor does it seem to matter very much. On the one hand, there being little historical or material foundation for the socialist project as an outcome of real conflicts and struggles generated by capitalism, that project has been reduced to a largely academic moral disputation, a rhetorical exercise designed to persuade. On the other hand, the terrain of moral contest has itself been severely restricted by a sanguine view of capitalism that narrows the ground on which socialism can claim superiority. It is, in any case, a simple matter of preference, and the choice appears to be of little moment.
It is not surprising, given the contingency and inconsequentiality of the choice between capitalism and socialism, that rcm tends to be obsessed with mathematical formulae and narrowly formalistic questions which have little connection with the real and urgent moralpolitical problems confronting humanity and the socialist movement today. Its constricted conception of the issues, and its casuistic formalism shaped to the narrow standards of a very particular specialized audience, are unlikely to produce anything particularly useful or politically effective. Still, it is perhaps early days and something more substantial may yet emerge from rcm efforts to construct a normative socialist theory. For the moment, if we are not simply to dismiss the moral argument out of hand as largely irrelevant to the realities of contemporary capitalism and the conditions of socialist struggle, we need at least to be persuaded that the moral theory, which in its most developed Roemerian form purports to reveal the exploitative foundations of capitalism, adds something to our understanding of capitalist exploitation, or rather, that it adds more than it subtracts.
Moral Argument v. Explanation?
Let us for the sake of argument accept that Roemer has indeed successfully revealed the moral principle on the basis of which capitalism stands condemned (Carling’s gloss on Roemer on pp. 41–2 of his first article may be taken as the most sympathetic account of the argument); and we can overlook the doubts cast by Elster on the usefulness of the concept of exploitation as a tool ‘for a more fine-grained investigation into moral theory’.  Elster, p. 229. The concern here will be with the consequences of the theory when it is given the benefit of the doubt. Here, Elster makes a striking point. He suggests that the very conceptual devices which constitute rcm as a moral-rhetorical strategy disable it as a mode of explanation.
The importance of exploitation in Marxism is twofold. First, the presence of exploitation in a society provides the outside observer with a ground for normative criticism. Exploitation is wrong; exploiters are morally condemnable; a society that tolerates or generates exploitation ought to be abolished. Secondly, exploitation can provide the exploited with a ground for taking individual or collective action against the system, and hence enters into the explanation of such action. When constructing a more elaborate theory of exploitation, one may face the problem that the normatively relevant concept is one that does not have much explanatory significance.  Ibid., p. 166.
Elaborating on this point later in the book, Elster writes:
the immediate transfer of surplus-value is not an important notion from the moral point of view. . . . Hence the moral and the explanatory aspects of Marxism diverge rather sharply at this point. It would have been theoretically satisfactory to argue that the grounds on which capitalism is to be condemned are also those that will motivate the struggle to abolish it. Marx, however, does not succeed in showing that this connection holds.  Ibid., p. 340.
Elster’s principal concern has to do with the role of class struggle in the socialist project: if the moral evil of capitalism lies in its unequal distribution of the relevant ‘resources’, while conflict and struggle are generated by a transfer of surplus which has no place in the moral argument, then the moral condemnation of capitalism has little to do with the motivations for struggles against it.
It needs to be said, first, that ‘classical’ Marxism never did make this kind of connection, in the form suggested by Elster. Certainly it is a distinctive characteristic of Marxism to insist on an organic connection between theory and practice; but this does not take the form of claiming that the ‘scientific’ analysis of capitalism can also yield a moral judgement which will motivate the revolutionary transformation. The argument is rather that the explanatory theory which exposes the workings of capitalism also expresses the self-emancipatory project of the working class by revealing the convergence of its interests and capacities with the conditions of a socialist transformation made possible by capitalism. This is not to deny Marx’s moral condemnation of capitalism; but it is precisely because he deliberately avoided the kind of connection suggested by Elster that there has been so much controversy about the very existence of a moral dimension in Marx’s account of capitalism.  For an excellent account of the debate surrounding these issues, and a persuasive argument that Marx himself did, if involuntarily, have a concept of justice, see Norman Geras, ‘The Controversy About Marx and Justice’, nlr 150, March/April 1985, pp. 47–85.
The force of Elster’s argument, however, does not derive from the claim that Roemer has broken a link on which the coherence of classical Marxism depended. It lies rather in the suggestion that Roemer’s moral argument itself has no explanatory power, and even that there is a contradiction between the requirements of the moral argument and the explanatory capacity of Marxist theory. Elster diverges from Roemer in a way which reveals an important truth about the Roemerian theory and about rcm generally, a truth which Roemer often hides from himself by stressing the automaticity of the links connecting the unequal distribution of assets (the moment of exploitation) to the generation of classes and in turn to the necessity of the struggle for socialism. Elster recognizes that the connections have been severed—not just modified in their degree of necessity but cleanly severed—by the exclusion of the transfer of surplus labour from the theory of exploitation. This affects not only our understanding of the revolutionary agency which will bring about socialism but also the whole Marxist explanation of history.
Elster’s answer, however, is not to reintroduce the transfer of surplus labour into the original account of exploitation, but rather to isolate the definition of exploitation as a normative concept while displacing the transfer of surplus labour to a different, explanatory plane. He suggests that without the ‘surplus labour theory’ there is much that Roemer cannot explain—in particular, about social conflict; but the conclusion he draws from this is simply that a theory of exploitation which might serve to establish a normative concept must be essentially devoid of explanatory power.
By itself, the separation of normative and explanatory arguments need not be disabling. It can be a legitimate exercise to construct a moral argument at an angle different from that required by, say, historical explanation. But problems do arise when there are outright contradictions, or when the premises of the moral argument require us to believe things which flout historical evidence or cloud our historical vision. The problems here do not flow simply from the decoupling of the moral and explanatory aspects of Marxism. The real difficulties lie in the explanatory implications of the moral argument itself—or rather, the explanatory weakness of rcm is traceable to the very first step in its moral argument, the point at which the moment of surplustransfer was detached from the unequal distribution of assets. For Roemer, the requirements of the moral theory were not—yet—visibly distinct from those of the explanatory model because both the moral and the causal aspects of exploitation had their origins in inequality and the apparent automaticity of its consequences. For Elster, the necessary consequence of the moral argument, which required isolating the moment of injustice by situating it in the original unequal distribution, is that moral and explanatory theories had to diverge.
But from then on—and this is the critical point—the separation remains in place not only for the analytic purpose of isolating the normative moment but as an intrinsic element in the explanatory model. In order to make his moral argument about exploitation without resorting to the labour theory of value, Roemer jettisons the whole apparatus which renders explanation possible, and thereafter seeks to reconstruct Marxist theory upon this hollow foundation. His vacuous account of history is the most notable result. Whether the fateful step was taken in order to meet the needs of an analytic theory of justice, or in order to satisfy the requirements of a game-theoretic explanatory model, or both together, the effects are the same. Either way, the theory of exploitation entails breaking the connections among the various ‘moments’ of capital—production, distribution, exchange, consumption—and destroying any conception of capitalism as a process, in which these analytically distinct moments are dynamically united. In other words, there has been a complete reversal of the theoretical advances which Marx so painstakingly laboured to establish in his critique of political economy. That this represents progress in our understanding of capitalism, and not a regression to the worst simplifications of pre-Marxist political economy, remains to be demonstrated.
Przeworski on Roemer: Exploitation and Class Struggle
Przeworski too recognizes the problem in Roemer’s argument—yet it is in his own attempt to find a solution that the weaknesses in the rcm paradigm are most starkly revealed. Roemer’s definition of exploitation, writes Przeworski, is superior to the ‘surplus value’ definition ‘not because it provides a better causal explanation’ but because it ‘makes clear “the ethical imperatives” of Marxian theory’.  Przeworski, p. 226. Indeed, he argues, the explanatory power of Roemer’s theory is so weak that it ‘does not establish any logical correspondence between exploitation and class struggle, that no such correspondence is here to establish, and that his historical assertions [about the necessity of the connections between inequality and exploitation, and between exploitation and class struggle—emw] are entirely rhetorical.’  Ibid., p. 227. There is, for Przeworski, no ‘tight correspondence’ of the kind proposed by Roemer between the original distribution of assets and its consequences in the inequality of income. The connection between the two is mediated and variable, subject, for example, to struggles between capital and labour at the point of production. Roemer simply factors out such mediations and thereby assumes out of existence ‘the problem of extracting labour out of labour-power’.  Ibid., p. 229. (Przeworski might have added that the rcm model in general does not accommodate the open-endedness, the incompleteness of the capitalist labour contract but proceeds as if the exchange between capital and labour were between two known quantities, a wage in exchange for a given service or product.) The automaticity of Roemer’s equation, Przeworski argues, does not allow, among other things, for the possibility that people will work more intensively for themselves than for others, and that the income of capitalists will depend on their success in extracting surplus labour from their reluctant workers. The problem, suggests Przeworski, is that Roemer has simply transported the assumptions of his original model of exploitation, where no exchange of labour takes place and where there is no problem of extracting labour, to an exploitative relation where there is a labour exchange and where the original assumptions are therefore no longer appropriate.
The rigid necessity of the connections in Roemer’s formula, between the distribution of assets and the distribution of income, leaves no logical place, argues Przeworski, for class struggle within a mode of production but only in the transition from one mode to another. This arbitrary automaticity allows Roemer to assume that no significant redistribution of wealth is possible under capitalism, given its original distribution of assets, and that workers are obliged to struggle for socialism if they want to improve their material conditions. Przeworski sets out to correct Roemer’s formulation in order to make room for class struggle within capitalism, to reconnect exploitation with class struggle and the process of capital accumulation, and above all to lay the foundation of an argument for socialism which does not depend on its presumed superiority in satisfying workers’ material interests.
There are some perceptive and telling arguments here, but it turns out that Przeworski is himself handicapped from the start by his own adherence to the rcm model. Indeed, he adheres to it in some respects more rigidly than does Roemer. The trouble with Roemer, apparently, is that he is not enough of a game-theorist, and that he has not introduced enough choice or contingency into the connection between exploitation and class struggle. But Przeworski’s faithfulness to the game-theoretical model leaves him with problems of his own. To Roemer’s rigid necessities, he can reply only with almost unlimited contingencies. Since for the purposes of the rational-choice model capitalism exists only as a ‘game’ between consenting adults, a game in which the stakes are no more compelling than a desire for ‘optimization’, there is no ‘logical place’ in his argument for capitalism as a compulsive and coercive system, a ruthless logic of process.
There is little room, first, for the compulsions which operate on capital, the imperatives of accumulation and self-expansion. We no longer have any notion of how to explain the relation between the ‘distribution of assets’ and the need for accumulation, the necessity of extracting surplus at the point of production and the contradictions and conflicts this necessarily occasions. All we have is a game in which capitalists have ‘assets’ which, if they wish, they can use to increase their wealth by extracting surplus from workers, as long as they succeed in overcoming whatever power of resistance the workers may have.
Second, there is little room for the fundamental coercion imposed on the worker from the beginning, before the ‘moment’ of labour-extraction and independent of the balance of power between any particular capitalist and any particular group of workers. This is the compulsion inherent in the initial disposition of property relations which systematically obliges some people to forfeit surplus labour to others simply in order to gain access to the means of labour itself.
If Roemer’s model is inadequate because of its failure to acknowledge the problem of extracting labour from labour-power, Przeworski’s is inadequate because of its failure to appreciate the coercive presuppositions of surplus extraction. If Roemer’s paradigm suffers from an inability to allow that people may work harder for themselves than for others, Przeworski’s is weakened by an inability to acknowledge the strength of the compulsion to intensify labour in cases where the obligation to labour for others is the pre-condition of access to the means of labour. If Roemer transports assumptions from one model to another where they do not apply, from an exploitative relation without a labour exchange to one in which there is such an exchange, an analogous transportation of assumptions occurs in Przeworski’s model: exploitative relations where access to the means of survival is conditional on the transfer of surplus labour are here analysed on the assumptions of a model where no such condition exists. If the explanatory weakness of Roemer’s model lies in its presumption of necessity, Przeworski’s fails because of its insistence on contingency.
In the guise of establishing connections between exploitation, class struggle and capital accumulation (connections which he says are missing in Marx as well as Roemer), Przeworski has effectively dissolved the systemic connections and compulsions long ago demonstrated by Marx. Capitalism appears at best as an inert ‘constraint’ on individual choices, and the constraint seems very weak, very permissive in the latitude of choices it allows. Przeworski’s argument is, in fact, based in large part on the assumption that a ‘major redistribution of wealth endowments’ is ‘feasible’ under capitalism, that class struggle can not only accelerate or retard accumulation but ‘transform’ it altogether while leaving the original distribution intact. This highly abstract proposition evinces an attitude to historical fact even more off-hand than Roemer’s.  Ibid., pp. 236–7. Przeworski here makes a truly extraordinary argument in support of his claims. To demonstrate that a ‘major redistribution’ within capitalism is historically no less feasible than socialism, he cites Henri Pirenne’s suggestion that ‘from the beginning of the Middle Ages to our time . . . for each period into which our economic history may be divided, there is a distinct and separate class of capitalists. In other words, the group of capitalists of a given epoch does not spring from the capitalist group of the preceding epoch. At every change of economic organization we find a breach of continuity . . . there are as many classes of capitalists as there are epochs in economic history.’ (Henri Pirenne, ‘The Stages of the Social History of Capitalism’, American Historical Reviewxix, 1914, pp. 494–5.) Quite apartfrom the flippant attitude to history and historical evidence revealed by this perfunctory reference to one short, outdated, and highly debatable article, in order to support a very large historical claim, Przeworski’s appeal to Pirenne on this point suggests a rather hazy understanding of capitalism. Pirenne is notoriously loose in his usage of ‘capitalism’ (though not uniquely so), and is generally inclined to imagine capitalism as already present in or coexisting with other social forms. The ‘capitalist’ is virtually anyone engaged in commerce for profit, any (urban) trader or merchant who, driven by the ‘love of gain’, uses movable property to amass more wealth, usually simply by buying cheap and selling dear. By these standards, the Renaissance Florentine merchant is a capitalist, but the English ‘improving’ landlord, orthe capitalist tenant farmer of the famous English ‘triad’, extracting surplus value from a wage labourer, is not. In the article in question, Pirenne simply outlines a series of different merchant types who, he argues, successively came to the fore in European economic history. What prevents him from pushing capitalism even further back in history, to ancient Greece or Rome, is not any conviction about the historical specificity of capitalism, but simply the fact that the evidence is insufficient. There is no sense in his argument of a specifically capitalist dynamic, a logic of accumulation essentially different from the age-old inclination to buy cheap and sell dear. It is, in any case, hard to see what support his argument lends to Przeworski’s contention about the feasibility of a major redistribution within capitalism, since Pirenne’s argument has nothing to do with a redistribution between exploiting and exploited classes. It is perhaps a measure of the imperatives imposed by the rational-choice model that Przeworski ends by contradicting the very convictions concerning the implacability of capitalism, and the limited scope it permits to reform, upon which he bases his critique of social democracy.
In the end, Roemer’s compulsive paradigm may be less misleading about the coercive realities of capitalism than is Przeworski’s permissive model—though the latter is more logically consistent with rcm premises, its conception of choice and the weak compulsion of ‘optimizing strategies’. Yet, though the consequences of Roemer’s procedure are masked in his own work by his insistence on the absolute necessity of everything that follows from unequal distribution, it cannot be said that his rcm critics are wrong to claim that the tight connections in Roemer’s equation are largely arbitrary and rhetorical; for once having made the first move in detaching the moments of capital, once having undermined the power of his theory to explain the coercive dynamic of capitalism, Roemer leaves no barriers to the conclusions drawn by Elster and Przeworski about the permissiveness of capitalism or the contingency of the relation between exploitation and class struggle.
If the moral theory of rcm is weak on its own terms, as a moral theory, because it lacks any purchase on the major moral and political issues of our time, it is doubly weak because it further undermines the power of the model to explain the conditions in which those issues must be contested. If rcm is to make good its claims as an improvement on conventional Marxism on any grounds, we have a right to expect a moral argument much more powerful than anything produced so far, not only to fill the gaps in the existing moral theory but to compensate for the explanatory sacrifices made on its behalf. The prospects, however, do not seem promising, for reasons that are inherent in the model itself. It seems unlikely that a moral theory capable of coming to grips with the realities of contemporary capitalism will come out of the constricted conception of the world and human experience contained in the rcm model. That model tends to be simplistically ‘economistic’ in its theoretical focus. Whatever their individual convictions about the complexity of human motivation, rcmists remain trapped in a theoretical paradigm which is blind to any motivating forces that cannot be reduced to the narrow terms of ‘market-rationality’, the calculation of ‘utilities’. It often seems that rcm, like the crudest Benthamite utilitarianism, can acknowledge only those emotions or beliefs which can be rendered in potentially quantifiable or economic terms—so that, for example, Elster can write of guilt or shame as ‘a utility fine’, and something like parental care for children might appear as a ‘transfer of utility’. The model is, after all, designed according to the specifications of bourgeois economics. In this respect, the rcm image of human nature has more in common with the ‘economic man’ of liberalism than with historical materialism. It is difficult to imagine that any profound moral insights will emerge from this banal and shrivelled homo economicus.
It is a telling comment on the weakness of Roemer’s claim to provide the essential moral critique of capitalism that Przeworski, the most explicitly political of the three rcmists, feels compelled in the end to step outside the paradigm to justify a preference for socialism. Capitalism, he argues, can meet any challenge on the economic terrain marked out by Roemer. If socialism is preferable at all, it must be because it permits ‘society as a whole to choose in a democratic way’ how to allocate its resources.  Przeworski, p. 238. The decisive moral criterion is not, then, equality but something like autonomy, or perhaps collective responsibility and community. Nothing in the rcm theory of exploitation has prepared us for this new moral standard.
But there is an even more curious anomaly. Roemer tells us that, according to historical materialism as he understands it, transformations of property relations are not caused by ideas about injustice and exploitation but by material conditions.  Roemer, Free to Lose, p. 124. Roemer does not present this account of historical materialism as contradicting his view that ‘it is perceptions and ideas about justice that are at the root of people’s support for or opposition to an economic system’ (p. 3), a view which is not inconsistent with a materialist account of historical transformations or even a materialist account of moral perceptions themselves. He does not dissociate himself from this proposition. Indeed, his adherence to technological determinism suggests a particularly deterministic interpretation of Marxism. What, then, is the point of reconstructing the whole of Marxist theory to satisfy the needs of an abstract moral argument? Why should rcm proceed as if moral persuasion were the principal agency of revolutionary change?
5.What Is the Politics of rcm?
The intellectual pay-off of Rational Choice Marxism appears to be strictly limited, and the costs may be far greater than the benefits. There remains the question of its political charge. A certain caution is required in the attempt to extract political implications from a theoretical tendency as abstract and formalistic as this. Its political significance may, in fact, lie precisely in a political amorphousness which makes it vulnerable to the vagaries of political fashion. One or two things can, however, be suggested about the political direction in which we are pointed by the theoretical assumptions of rcm.
If one were simply to list the principal features of the rcm model, the result would be something very like a caricature of Anglo-American liberalism as it has evolved since the seventeenth century: methodological individualism; ‘analytic’ method; ahistoricism (which is not necessarily incompatible with technological determinism or its functional equivalent and frequent corollary, a conception of history as the triumph of ‘commercial society’); class conceived as income stratification; a preoccupation with market relations as distinct from production relations; an ‘economic’ model of human nature. This theoretical constellation could represent a rough sketch of the Anglophone liberal mind-set with its typical symbiosis of liberal ideology and British empiricism, in which a reductionist focus on human nature has been associated with a formalistic tradition of analytic philosophy. The striking resemblance between rcm and this liberal-empiricist idealtype does not, of course, guarantee that all, or any, rcmists must subscribe to the relevant political doctrines; but the analogy is at least suggestive.
At the same time, there is another, at first glance opposing, tradition to which rcm has certain striking affinites—utopian socialism: a detachment of the ethical ideal of socialism from the historical conditions of its realization; a distributional theory of exploitation which locates the moment of injustice in the sphere of circulation and exchange; a ‘one-sided’ presentation of capitalism which abstracts the ‘free’ (if ‘unfair’) exchange between capital and labour from its ‘presuppositions’, thereby conjuring away the barriers between capitalism and socialism by implicitly constructing a continuum from capitalist to socialist ‘freedom and equality’. This is a tradition about which Marx had a great deal to say, much of it uncannily prophetic of rcm, culminating in the following assessment:
It is forgotten, on the one side, that the presupposition of exchange value, as the objective basis of the whole of the system of production, already in itself implies compulsion over the individual, since his immediate product is not a product for him, but only becomes such in the social process, and since it must take on this general but nevertheless external form; and that the individual has an existence only as a producer of exchange value, hence that the whole negation of his natural existence is already implied, that he is therefore entirely determined by society; that this further presupposes a division of labour etc., in which the individual is already posited in relations other than that of mere exchanger, etc. That therefore this presupposition by no means arises either out of the individual’s will or out of the immediate nature of the individual, but that it is, rather, historical, and posits the individual as already determined by society. It is forgotten, on the other side, that these higher forms, in which exchange, or the relations of production which realize themselves in it, are now posited, do not by any means stand still in this simple form where the highest distinction which occurs is a formal and hence irrelevant one. What is overlooked, finally, is that already the simple forms of exchange value and of money latently contain the opposition between labour and capital etc. Thus, what all this wisdom comes down to is the attempt to stick fast at the simplest economic relations, which, conceived by themselves, are pure abstractions; but these relationships are, in reality, mediated by the deepest antithesis, and represent only one side, in which the full expression of the antithesis is obscured.  Marx, Grundrisse, pp. 247–8.
This is Marx’s account of the capital relation as it appears to those ‘foolish’ French socialists ‘who want to depict socialism as the realization of the ideals of bourgeois society’. And here perhaps is the crux of the matter. If the theoretical apparatus of rcm has any specific political implications, they too may lie in the conflation of capitalism and socialism which inheres in the very structure of the argument and which would account for its joint affinities to Anglo-American liberalism and French utopian socialism.
Conflating Capitalism and Socialism
Carling’s text on Rational Choice Marxism illustrates how the theoretical imperatives of rcm, and its rhetorical or ethical requirements in particular, exert a political pressure of their own. The essence of his argument has to do with how and why it might be philosophically legitimate to regard capitalism as exploitative, and hence unjust, despite the freedom and equality inherent in it. At first, this procedure appears to be a rhetorical strategy designed to meet defenders of capitalism on their own terrain. Carling suggests, however, that these rhetorical devices contain an important truth and that we should—as he would argue Roemer does—take seriously the claims to freedom and equality contained in their account of capitalist relations.
The extent to which Carling’s argument here depends upon a ‘onesided’ view of capitalism is nowhere more strikingly illustrated than in his suggestion that ‘the market-place really is a free space: what is constitutionally unfree is the State-place, which upholds property, and the work-place, where those without property find it is in their best interest to labour for their daily bread.’  Carling, I, p. 36. (Note the rather coy formulation according to which propertyless workers ‘find it is in their best interest’ to labour for capital.) Carling thus completely misses the essence of capitalist unfreedom on two principal counts—by masking the necessity of the ‘choice’ between exploitation and survival rooted in the structure of capitalist social-property relations, and the ultimate compulsion of the capitalist market, the universal subjection of all human beings to its sovereignty summed up in Marx’s ‘fetishism of commodities’. He thereby evacuates the ground for any adequate statement of the socialist project. What he concludes from his analysis of capitalist freedom and equality is simply that we should regard socialism not as something qualitatively different from capitalism but rather as more of the same, more of the freedom and equality already inherent in capitalism.
In Carving’s account of rcm there appears to be an explicit warrant for drawing political-strategic conclusions from the rhetorical merger of capitalism and socialism. But even short of that, the tendency to conflate capitalism and socialism, which is inherent in the rcm project, has political implications. If socialism is simply a quantitative improvement, an extension of capitalist freedom and equality, the passage from one to the other is likely to be relatively smooth and non-antagonistic—with all the implications this perception has for the marginalization of class politics. But even if the conflation is merely a rhetorical device, then at the very least it guarantees the vacuity of rcm as a guide to the strategy, because its moral-rhetorical value depends precisely on ignoring the critical barriers—such as class antagonisms—standing in the way of a smooth transition from one social form to the other.
There is, however, a significant paradox here—a paradox that belongs to the essence of every utopian socialism. On the one hand, the conflation of capitalism and socialism appears to suggest that the latter will grow directly, and more or less smoothly, out of the former. On the other hand, that same conflation is accompanied by, and indeed has as its principal corollary, a detachment of the socialist project from any historical foundation in the actual conditions of capitalism. Socialism grows out of capitalism not in the sense that the one creates the structural and historical conditions which make the other possible, the contradictions which place socialism on the historical agenda, and the agencies capable of carrying out the socialist project, but rather in the sense that socialism is a ‘realization of capitalist ideals’. This juxtaposition of apparently contradictory ideas helps to account for the curious blend of optimism and pessimism which so often characterizes this kind of socialism: the transition to socialism will come about smoothly, without radical breaks or antagonistic encounters, but it is a very long way off—so long that it has receded into invisibility; hence the task of socialists is to ‘humanize’ capitalism. It may also help to explain the ambivalent quietism, even cynicism, of an Adam Przeworski, in which a hard-headed scepticism about the prospects of socialism exists in uneasy tandem with a trenchant critique of social democracy.
The rcm analysis of capitalism—its detachment of the ‘moments’ of capital, its theoretical annihilation of the capitalist system and capitalistprocess, the effects of these procedures in dissociating class and class struggle from the logic of the capitalist process and rendering them contingent—severs the socialist project from its historical and political roots in the conditions of capitalism. The feasibility of the socialist project, and the possibility of mobilizing forces to carry it out, are no longer historical and political questions but simply rhetorical ones, having to do with the discursive conditions for persuading ‘free and rational’ individuals to ‘choose’ socialism—conditions that have little to do with the processes by which political movements are actually formed.
The ‘tough-minded’ scepticism about the feasibility of socialism expressed by rcmists like Adam Przeworski, in other words, has less to do with a hard-headed assessment of capitalist realities than with an abdication of the ground on which any such assessment could be made. If there is a difference between rcm and utopian socialism in these respects, it is not that the former provides a more effective analysis of the historical conditions for the realization of socialist ideals but rather that it offers a less effective, because more constricted and less passionate, moral vision.
The Convergence of Post-Marxist Theories
There is, it is true, a range of political positions among exponents of the rcm ‘paradigm’, from Elster and (now) Przeworski on the right to Roemer further left. But if the political logic of the theoretical model is too weak to produce any single political outcome—apart from an inevitable strategic vacuity—the developmental trajectory of the trend as a whole is instructive. There has been a visible shift from rcm’s original efforts to set Marxism on a more rigorous analytic footing, against attacks from the right, to what Elster has described as an ‘unstated consensus’ among its practitioners that leaves ‘probably not a single tenet of classical Marxism’ intact.  Elster, p. xiv. As ‘classical Marxism’ has given way to neo-classical economics, game theory, methodological individualism, and neo-contractarian philosophy, as rcm has set itself the task of ‘making sense’ (in Elster’s phrase) of Marxist nonsense, so too the political commitment to socialist values has been tempered by a new ‘realism’ about the prospects of socialism. The cynical quietism evinced in the latest work of Adam Przeworski suggests the direction in which the logic of the trend is moving.
In some respects, this trajectory has much in common with another major theoretical tendency in contemporary Marxism, from Althusserianism to ‘post-Marxism’, which also began as an effort to reestablish ‘rigour’ in Marxist theory and has ended for many in a general repudiation of Marxism in theory and practice. In the latter case, the theoretical evolution of the trend is traceable to immediate historical coordinates, and especially to the life-cycle of the European Communist movement, or one of its significant off-shoots, as it wound its tortuous way from flirtations with Maoism through Eurocommunism and its current fatal disarray. rcm has not been similarly anchored to a political movement of this—or any other—kind, though it has been shaped by the general move to the right in its own geo-political backyard. But if its academic detachment has acted as a safeguard against some varieties of dogma, the absence of roots in the labour movement has also made it more susceptible to other dangers—not the least of which are the pressures of the academy, the attractions of current academic fashion, the professional requirements of an academic career, standards of judgment and a sense of proportion deriving not from the political arena but from the senior common room (or its American and/or Scandinavian equivalent), where the principal adversary is likely to be a professional neo-classical economist.
We may now be observing a curious convergence between two apparently antithetical tendencies, the super-rationalism of Rational Choice Marxism and post-structuralist irrationalism. rcm’s abstractions, so typical of analytic philosophies which pride themselves on their empyreal detachment from the uncertain flux of historical process, paradoxically come together here with the irrationalist dissolution of history from the opposite direction. Both are impelled toward a politics detached from the anchor of history, as game-theoretic choices join post-modern contingency in a contradictory amalgam of political voluntarism, where rhetoric and discourse are the agencies of social change, and a cynical defeatism, where every radical programme of change is doomed to failure.
[*] I am very grateful to Perry Anderson, Robert Brenner, Diane Elson, Norman Geras and Neal Wood for their comments and suggestions.
 Alan Carling, ‘Rational Choice Marxism’, nlr 160, November/December 1986, p. 55. (Hereafter, this article will be cited as Carling I.)
 Carling I, pp. 26–7.
 Carling I, p. 52.
 Carling, ‘Liberty, Equality, Community’, nlr 171, September/October 1988, p. 95, (Hereafter, Carling II.)
 Andrew Levine, Elliott Sober, and Erik Olin Wright, ‘Marxism and Methodological Individualism’, nlr 162, March/April 1987, pp. 67–84. For an even more recent version of Wright’s theory of class, see Wright et al., The Debate on Classes, Verso, orthcoming.
 John Roemer, ‘Methodological Individualism and Deductive Marxism’, Theory and Society 11 (1982), p. 513.
 Jon Elster, Making Sense of Marx, Cambridge 1985, p. 7.
 According to this criterion, a ‘coalition of agents’ is exploited if it would be better off withdrawing from, rather than remaining within, the ‘game’ with its per capita share of whatever assets are relevant to the mode of exploitation in question.
 Roemer, A General Theory of Exploitation and Class, Cambridge, Mass. 1982, pp. 20–21.
 Roemer, Free to Lose, London 1988, p. 136.
 Roemer, General Theory, p. 20.
 This is certainly not to deny that Roemer’s theory of exploitation is open to criticism even on its own economic terms. See, for example, Ronald A. Kieve, ‘From Necessary Illusion to Rational Choice? A Critique of Neo-Marxist Rational-Choice Theory’, Theory and Society 15 (1986), esp. pp. 558–565. My own argument has more to do with evaluating the utility of this theory even if it succeeds on its own terms.
 Elster, p. 515.
 Adam Przeworski, Capitalism and Social Democracy, Cambridge 1986, p. 97.
 Przeworski, p. 96.
 Przeworski, p. 97.
 Przeworski, p. 97.
 Elster, p. 13.
 For a discussion of this point, see Peter Meiksins, in Wright et al., The Debate on Classes.
 Elster, pp. 203, 335–42.
 Elster, p. 12, Marx, Grundrisse, Harmondsworth 1973, p. 283.
 Elster, p. II.
 Grundrisse, p. 284.
 Elster, pp. 310–17, 180 ff.
 Levine, Sober and Wright, p. 80.
 Levine, Sober and Wright, p. 82.
 Roemer, Free to Lose, p. 172.
 Carling I, p. 30.
 See Elster, pp. 338 ff.
 In ‘Methodological Individualism and Deductive Marxism’, for example, Roemer seems to be arguing that methodological individualism of the type advocated by Elster is essential to the explanation of historical transformations and superior to conventional Marxism in this respect.
 Carling II, p. 95. The relevant reference for G.A. Cohen is Marx’s Theory of History: A Defence, Oxford 1978. The principal references for Robert Brenner are ‘The Social Basis of Economic Development’, in J. Roemer, ed., Analytical Marxism, Cambridge 1985; ‘The Origins of Capitalist Development: A Critique of Neo-Smithian Marxism’, nlr 104, July/August 1977, pp. 25–92; and The Brenner Debate: Agrarian Class Structure and Economic Development in Pre-Industrial Europe, T.H. Aston and C.H.E. Philpin eds., Cambridge 1985.
 Carling II, p. 95, and Roemer, Free to Lose, pp. 123–4.
 Carling II, p. 95.
 Roemer, Free to Lose, p. 108.
 Roemer, Free to Lose, p. 126.
 Roemer, Free to Lose, p. 6. Emphasis in the original. Given rcm’s need to lower the stakes in the game of class, it may be significant that Roemer chooses to emphasize the progressive socialization of property as the elimination of exploitative forms, in contrast to Marx’s focus on history as the progressive separation of direct producers from the conditions of their labour (this is, for example, the basic premise of the Grundrisse).
 Roemer, Free to Lose, p. 115.
 Roemer, Free to Lose, p. 124.
 Roemer, Free to Lose, p. 123.
 Roemer, General Theory, pp. 270–71.
 Roemer, Free to Lose, p. 136.
 Robert Brenner, ‘Bourgeois Revolution and Transition to Capitalism’, in The First Modern Society, A.L. Beier et al., eds., Cambridge 1989, p. 280.
 Brenner, ‘Bourgeois Revolution’, p. 287.
 Brenner, ‘Bourgeois Revolution’, p. 293.
 Brenner’s contribution to the collection on Analytical Marxism edited by Roemer, which presents his account of the transition in a schematic form, argues that the conventional explanation of the development of capitalism, based largely on Adam Smith, assumes precisely the extraordinary phenomenon that needs to be explained; that property relations must be understood as ‘relations of reproduction’; that precapitalist economies have their own logic and ‘solidity’, which are in effect denied by the conventional view; that capitalist development is a more historically limited and specific occurrence than is allowed by theories which attribute it to some universal law of technical progress; and that the history of the transition cannot be explained by assuming that there is a necessary correspondence between the self-interested actions of individual actors and the requirements of economic growth. In these respects, his argument runs directly counter to Roemer’s basic assumptions, and to Cohen’s technological determinism. It is worth adding that his schematic argument in this volume, in contrast to the contributions of Rational Choice Marxists, is grounded in historical work and is based on the premise that the work of historical explanation needed to be done in advance of the analytic presentation.
 Rodney Hilton, The Transition Debate, Verso, London, 1976. See also Hilton, ‘Towns in English Feudal Society’, in Class Conflict and the Crisis of Feudalism: Essays in Medieval Social History, London 1985.
 It was Marx himself who insisted that capitalism is unique in its drive to revolutionize productive forces, while other modes of production have tended to conserve existing forces, and that ‘petrification’ may have been the rule rather than the exception. See, for example, Capital I, Moscow 1971, pp. 456–7. A similar view appears even in the Communist Manifesto, which in other respects still adheres to the early, uncritical theory of history. I have criticized Cohen’s technological determinism elsewhere, in ‘The Separation of the Economic and the Political in Capitalism’, nlr 127, pp. 70–74, and in a more general discussion of technological determinism in ‘Marxism and the Course of History’, nlr 149, pp. 95–107.
 Elster, p. 54.
 Marx, Grundrisse, p. 105.
 See, for example, Roemer, Free to Lose, p. 151. The weakness of the barrier erected by rcm against right-wing triumphalism is illustrated by John Gray’s review of Free to Lose in the Times Literary Supplement, 24 February–2 March 1989. Roemer is here castigated for adopting the worst features of neo-classical economics, its complete abstraction and detachment from the real world of market processes—though in Gray’s account, of course, the real world triumphantly vindicates capitalism and decisively demonstrates the inferiority of socialism. Nothing in Roemer’s argument equips him to withstand this onslaught.
 Przeworski, pp. 237–8.
 Elster, p. 229.
 Ibid., p. 166.
 Ibid., p. 340.
 For an excellent account of the debate surrounding these issues, and a persuasive argument that Marx himself did, if involuntarily, have a concept of justice, see Norman Geras, ‘The Controversy About Marx and Justice’, nlr 150, March/April 1985, pp. 47–85.
 Przeworski, p. 226.
 Ibid., p. 227.
 Ibid., p. 229.
 Ibid., pp. 236–7. Przeworski here makes a truly extraordinary argument in support of his claims. To demonstrate that a ‘major redistribution’ within capitalism is historically no less feasible than socialism, he cites Henri Pirenne’s suggestion that ‘from the beginning of the Middle Ages to our time . . . for each period into which our economic history may be divided, there is a distinct and separate class of capitalists. In other words, the group of capitalists of a given epoch does not spring from the capitalist group of the preceding epoch. At every change of economic organization we find a breach of continuity . . . there are as many classes of capitalists as there are epochs in economic history.’ (Henri Pirenne, ‘The Stages of the Social History of Capitalism’, American Historical Reviewxix, 1914, pp. 494–5.) Quite apartfrom the flippant attitude to history and historical evidence revealed by this perfunctory reference to one short, outdated, and highly debatable article, in order to support a very large historical claim, Przeworski’s appeal to Pirenne on this point suggests a rather hazy understanding of capitalism. Pirenne is notoriously loose in his usage of ‘capitalism’ (though not uniquely so), and is generally inclined to imagine capitalism as already present in or coexisting with other social forms. The ‘capitalist’ is virtually anyone engaged in commerce for profit, any (urban) trader or merchant who, driven by the ‘love of gain’, uses movable property to amass more wealth, usually simply by buying cheap and selling dear. By these standards, the Renaissance Florentine merchant is a capitalist, but the English ‘improving’ landlord, orthe capitalist tenant farmer of the famous English ‘triad’, extracting surplus value from a wage labourer, is not. In the article in question, Pirenne simply outlines a series of different merchant types who, he argues, successively came to the fore in European economic history. What prevents him from pushing capitalism even further back in history, to ancient Greece or Rome, is not any conviction about the historical specificity of capitalism, but simply the fact that the evidence is insufficient. There is no sense in his argument of a specifically capitalist dynamic, a logic of accumulation essentially different from the age-old inclination to buy cheap and sell dear. It is, in any case, hard to see what support his argument lends to Przeworski’s contention about the feasibility of a major redistribution within capitalism, since Pirenne’s argument has nothing to do with a redistribution between exploiting and exploited classes.
 Przeworski, p. 238.
 Roemer, Free to Lose, p. 124. Roemer does not present this account of historical materialism as contradicting his view that ‘it is perceptions and ideas about justice that are at the root of people’s support for or opposition to an economic system’ (p. 3), a view which is not inconsistent with a materialist account of historical transformations or even a materialist account of moral perceptions themselves.
 Marx, Grundrisse, pp. 247–8.
 Carling, I, p. 36.
 Elster, p. xiv.